Citation
Policy actor beliefs and behaviors in contentious policy debates

Material Information

Title:
Policy actor beliefs and behaviors in contentious policy debates examining policy actors within the statewide, fracking subsytems of Colorado, Texas, and New York
Creator:
Gallaher, Samuel Ballou ( author )
Place of Publication:
Denver, Colo.
Publisher:
University of Colorado Denver
Publication Date:
Language:
English
Physical Description:
1 electronic file (265 pages) : ;

Subjects

Subjects / Keywords:
Hydraulic fracturing -- Government policy -- Colorado ( lcsh )
Hydraulic fracturing -- Government policy -- Texas ( lcsh )
Hydraulic fracturing -- Government policy -- New York (State) ( lcsh )
Texas ( fast )
New York (State) ( fast )
Colorado ( fast )
Genre:
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )

Notes

Review:
The goal of this dissertation is to address three areas in the policy process literature that require clarification. First, it examines how a policy actor's deep and policy core beliefs translate into secondary beliefs. To do so, the research models the effect of an individual's view of government in daily life and their policy belief towards fracking on their secondary belief of which level of government should regulate an issue. Second, the research examines how a policy actor's policy core beliefs affect a behavior called venue shopping. The research asks how policy actors' belief towards the policy status quo affects their shopping activity level, and how their beliefs toward decision makers influence venue selection. Third, the research examines local governmental representatives as policy actors in a state-level policy subsystem. Policy process research identifies local government representatives within advocacy coalitions, but little is known about how local governmental actors compare to other advocates in the coalitions. The research uses the Advocacy Coalition Framework (ACF) as an analytical and theoretical foundation and applies other policy process and organizational theories to inform its hypotheses. I use multiple quantitative data modeling techniques to explore each question. Data for the research is from original surveys of policy actors in state-level hydraulic fracturing subsystems in Colorado, Texas, and New York. Findings indicate policy actors' deep core and policy core beliefs significantly influence their secondary beliefs. However, deep core beliefs have a greater effect on secondary beliefs related to more abstract issues, such as air quality, and less on more concrete issues, such as the distance a well should be from other structures. The venue shopping models indicate policy actors who oppose the policy status quo shop more venues than those who align with the status quo. Additionally, the strongest indicator of which venue a policy actor shops is not their beliefs toward the decision makers, but their other shopping choices. Finally, analyses show local governments are a unique group within and across coalitions because of their network relationships and they align with one another on a set of policy core beliefs, but are also divided among pro and anti-fracking coalitions on other policy core beliefs. Overall, this dissertation shows the ACF provides a theoretical and analytical frame to examine policy actor beliefs and behavior, but additional theories and sub-groupings of policy actors are needed to explain nuances in policy actor dynamics.
Bibliography:
Includes bibliographical references.
System Details:
System requirements: Adobe Reader.
Statement of Responsibility:
by Samuel Ballou Gallaher.

Record Information

Source Institution:
University of Colorado Denver
Holding Location:
Auraria Library
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
on10022 ( NOTIS )
1002218902 ( OCLC )
on1002218902
Classification:
LD1193.P86 2017d G35 ( lcc )

Downloads

This item has the following downloads:


Full Text
POLICY ACTOR BELIEFS AND BEHAVIORS IN CONTENTIOUS POLICY DEBATES:
EXAMINING POLICY ACTORS WITHIN THE STATEWIDE, FRACKING SUBSYSTEMS OF COLORADO, TEXAS, AND NEW YORK.
by
SAMUEL BALLOU GALLAHER
B.S., Oregon State University, 2004
M.P.A, University of Colorado Denver, 2011
A thesis submitted to the Faculty of the Graduate School of the University of Colorado in partial fulfillment Of the requirements for the degree of Doctor of Philosophy Public Affairs Program
2017


2017
SAMUEL BALLOU GALLAHER ALL RIGHTS RESERVED
ii


This thesis for the Doctor of Philosophy degree by
Samuel Ballou Gallaher has been approved for the Public Affairs Program by
Tanya Heikkila, Chair Christopher Weible, Chair Benoy Jacob Rick Feiock
Date: July 29, 2017


Gallaher, Samuel Ballou (Ph.D., Public Affairs)
Policy Actor Beliefs and Behavior in Contentious Policy Debates: Examining Policy Actors Within the Statewide, Fracking Subsystems of Colorado, Texas, and New York.
Thesis directed by Professor Tanya Heikkila mid Professor Christopher Weible
ABSTRACT
The goal of this dissertation is to address three areas in the policy process literature that require clarification. First, it examines how a policy actors deep mid policy core beliefs translate into secondary beliefs. To do so, the research models the effect of an individuals view of government in daily life mid their policy belief towards fracking on their secondary belief of which level of government should regulate mi issue. Second, the research examines how a policy actors policy core beliefs affect a behavior called venue shopping. The research asks how policy actors belief towards the policy status quo affects their shopping activity level, mid how their beliefs toward decision makers influence venue selection. Third, the research examines local governmental representatives as policy actors in a state-level policy subsystem. Policy process research identifies local government representatives within advocacy coalitions, but little is known about how local governmental actors compare to other advocates in the coalitions. The research uses the Advocacy Coalition Framework (ACF) as an analytical and theoretical foundation and applies other policy process mid organizational theories to inform its hypotheses. I use multiple quantitative data modeling techniques to explore each question. Data for the research is from original surveys of policy actors in state-level hydraulic fracturing subsystems in Colorado, Texas, and New York. Findings indicate policy actors deep core and policy core beliefs significantly influence their
IV


secondary beliefs. However, deep core beliefs have a greater effect on secondary beliefs related to more abstract issues, such as air quality, mid less on more concrete issues, such as the distance a well should be from other structures. The venue shopping models indicate policy actors who oppose the policy status quo shop more venues than those who align with the status quo. Additionally, the strongest indicator of which venue a policy actor shops is not their beliefs toward the decision makers, but their other shopping choices. Finally, analyses show local governments are a unique group within and across coalitions because of their network relationships mid they align with one another on a set of policy core beliefs, but are also divided among pro and anti-fracking coalitions on other policy core beliefs. Overall, this dissertation shows the ACF provides a theoretical and analytical frame to examine policy actor beliefs and behavior, but additional theories and sub-groupings of policy actors are needed to explain nuances in policy actor dynamics.
The form mid content of this abstract are approved. I recommend its publication
Approved: Tanya Heikkila mid Christopher Weible
v


For Caleb. He never gave up, so neither did I.
For Laura. Thank you for your unending love and support and patience. I love you.
For Mom and Dad.
VI


ACKNOWLEDGEMENTS
The success of this research and dissertation and the completion of the program ingeneral is due to a vast amount of support. First mid foremost, my co-advisors, PFs, and mentors, Tanya Heikkila mid Chris Weible. Thank you for your all you did to find resources and bring me into the world of academic research through conferences, presentations, roundtables, and workshops. Your approach to education and transformed me from an inquisitive individual to a thoughtful social scientist. I also wish to acknowledge your contributions to this dissertation. You both guided the research from beginning to end, including leading the development of the surveys used, involved discussions on the methods used and countless edits on each chapter. While I may not have always received edits and suggestions with a smile, once the wrangling of thoughts and words was done, the product was in better shape than before. Chris. Thank you for pushing me to become a better writer. Tanya. Thank you for helping me see that I am a c duplicator and, when appropriate, seek the path of the simplifier. Thank you.
Thank you to Rick Feiock and Benoy Jacob for your efforts on my dissertation committee. Your thoughtful comments and questions pushed me outside of my theoretical and interpretive comfort zone.
Thank you to my fellow cohort and WOPPR-ites mid, visiting scholar friend Jarkko. While we struggled in our own ways, and often in the isolation of our own research mid work, we formed friendships mid, on those wonderful occasions, got to relax with a good drink. Thank huge thank you to my fellow WOP PR researchers on the Sloan Grant. Jon Pierce, Jennifer Kagan, Ben Blair, our weekly meetings with Chris and Tanya were a highlight. I wont forget the constant Battlestar Gallatica inspired snickering, ceaseless
vii


survey editing, and just good of tomfoolery. Jennifer, I also wish to thank you for your help in collecting state-level regulatory information years ago. That information became a deep well to draw upon when writing my research setting sections. A special thanks to David Carter. I am thankful for our many fine conversations and that I learned how to annoy you. Your unrelenting work ethic and dedication to the craft was an inspiration.
Thank you to the School of Public Affairs and the staff and faculty. The first day I entered the school, a few weeks prior to starting the MPA program, I met with Richard Stillman mid shortly thereafter, Peter deLeon. Richard, your enthusiasm for unabashed learning and intellectual exploration was truly inspiring. Also, once you learned I hoped to find work as a research assistant you promptly introduced me to Peter. A meeting which I thought was a just a friendly introduction. To my surprise, by the time I had made my way back to the elevators, Peter handed me a job description to work for him and Chris on the Policy Studies Journal. And like that I was a graduate assistant. Who knows what would have happened if not for your way of supporting students? Thank you also to Malcom Goggin, Brian Gerber, Todd Ely, and Benoy Jacob. I am grateful for your willingness work with me and let me experience new mid exciting research and methods. Thank you to Rob Drouillard for your friendliness and ensuring we had much needed access to software and files, while working at school, home, or in Ecuador. Thank you to Dawn Savage. I cant believe the number of deadlines you kept me from missing and forms you reminded me to fill out. Thanks for your open door mid cheerful conversation!
Thank you to my friends and family who saw me less mid supported me more over the last six years. While I missed a lot, the game nights, camping trips, hut trips, dinners, gatherings, bachelor parties, weddings, and holidays that I made kept me grounded and sane.
viii


Preston, Josh, Mikey, Heather, Ian, Odele, Andy, Jill, Morgan, Luke, Ben, Sarah, Gavin, Curtis, Kara, Heather, Robert, Clif, Alex, Mike, Eva, Brian, Ingrid, Matt, Amanda, Joel, Sarah, Danielle, Nick, Karen, Nate, Kerry, David, Sarah, David, Val, Jeremy, Sarah, Amy, Katie, Alex, Tomoko, Marcus, Patricia, Laura. You are my village. Thank you, Mom and Dad, for never doubting my success, helping with food and plane tickets, and the gazillion questions every time I called. Thank you, David and Sarah, for your friendship and opening your home to me, and soon thereafter, to Laura. It is not always the case you get to call roommates friends.
Of course, I thank you Laura. Soon youll have to get to know me as a normal nongraduate-student human being. If your kindness, laughter, patience, and love maintain, Im sure well be fine.
IX


TABLE OF CONTENTS
CHAPTER
I. INTRODUCTION...........................................................1
Background.............................................................1
Research Setting: Statewide Hydraulic Fracturing (Fracking) Debates....4
Literature Review: The Advocacy Coalition Framework....................8
Chapter Introduction................................................. 17
II. TRANSLATING BELIEFS: HOW POLICY ACTORS DEEP AND POLICY CORE
BELIEFS AFFECT THEIR SECONDARY BELIEFS................................23
Introduction: Who Regulates...........................................23
Contributions and Map of the Paper....................................26
Theoretical Arguments and Hypotheses Development......................27
Research Setting......................................................33
Methods...............................................................42
Analysis mid Results..................................................50
Conclusion............................................................71
Discussion mid Limitations............................................74
III. POLICY ACTORS VENUE SHOPPING PATTERNS DURING NEW YORKS
FRACKING DEBATES......................................................78
Introduction: Venue Shopping..........................................78
Theoretical Foundations...............................................82
x


Hypotheses Development..................................................84
Research Setting........................................................94
Methods.................................................................96
Analysis mid Results...................................................102
Conclusions and Limitations............................................117
IV. COMPARING THE BELIEFS OF LOCAL GOVERNMENTAL POLICY ACTORS TO THEIR INTEREST GROUP ALLIES................................................121
Introduction: Local Governments as Policy Actors.......................121
Theoretical arguments and Hypothesis Development.......................123
Research Setting.......................................................128
Methods................................................................130
Analysis mid Results...................................................134
Conclusions and Limitations............................................148
V. COMPARING THE RESOURCES, NETWORKS, AND POLITICAL ACTIVITIES OF LOCAL GOVERNMENTAL POLICY ACTORS TO THEIR INTEREST
GROUP ALLIES..............................................................153
Introduction: Local Governments as Policy Actors..........................153
Theoretical Arguments and Expectations....................................157
Research Setting..........................................................166
Methods...................................................................167
XI


Analysis mid Results...................................................172
Conclusions and Limitations............................................186
VI. CONCLUSIONS.............................................................192
Summary of work........................................................192
Contributions..........................................................197
Future Research........................................................204
REFERENCES....................................................................206
APPENDIX......................................................................226
A. Chapter 2..............................................................226
Descriptive Statistics.................................................226
Hypothesis 1 tabular data..............................................228
Margins For Interaction Between Policy Preference And State............229
B. Chapter 3..............................................................233
Descriptive statistics for RQ1.........................................233
Descriptive statistics for RQ2.........................................234
Local Level Action.....................................................236
Alternative Model for RQ1..............................................237
C. Chapter 4..............................................................238
Respondent descriptive on policy position and org type.................238
HI: ANOVA..............................................................238
xii


HI: Full Fisher-Hayter Table..............................................239
HI: Correspondence Analysis...............................................240
H2: Difference in Means mid Extremism.....................................241
D. Chapter 5.................................................................244
Factor Analyses of Resources..............................................244
HI: Comparison of Resources...............................................246
H2: Primary and Secondary Activities......................................248
H2: OLS Progression of primary and secondary activities...................249
Network Size..............................................................251
Network pattern MCA.......................................................252
xiii


CHAPTER I
INTRODUCTION
Background
A governments policy decisions are powerful because they activate their ability to coerce actions of and redistribute resources to individuals and organizations. The Policy process is one way in which scholars describe and explain how governments and interest groups make and change policies. Depending on a researchers scale mid unit of inquiry, the policy process can be linear or something quite unlike a process. For example, a single policy may follow a general path from inception through implementation. However, when the researcher examines bundles of policies surrounding a topic, the policy process can be something with neither a beginning nor an end, only an evolution. In either case, the policy process is a messy construct full of individuals, acting over a landscape defined by rules, mid in an environment known for disruptions that come in the form of new and unexpected information. As policy scholars unpack the policy processes around a topic, a few of the big questions asked are who was involved in the process?, why did the outcome occur?, and how did the outcome occur? Two of the key elements used to explain those big questions are the beliefs and the behaviors of the individuals involved in the policy process. Indeed, while a multitude of institutions (e.g., rules, laws, or norms) govern the policy process, the policy process is also a social process in which individuals with strong belief mid, cognitive and physical limitations are synthesizing information and making decisions.
Policy scholars who examine policy actor beliefs mid behaviors often do so in the context of contentious debates. This is because contentious policy debates provide a research setting with rich variation in both belief mid behavior to explore and test hypotheses. Eight
1


defining characteristies of contentious debates highlight these points. First, contentious policy debates involve multiple advocacy groups working collectively based on shared interests for or against the policy or policies in contention (Tilly & Tarrow, 2007; Sabatier, 1988). Second, advocacy groups, defined as coalitions, consist of a multiplicity of policy actors from local, state, and federal levels of government, nonprofit mid for-profit organizations, media, and the scientific community (Sabatier, 1988). Third, the coalitions of policy actors engage in a wide variety of activities in mid out of the policy making venues (Tilly & Tarrow, 2007; Sabatier & Jenkins-Smith 1993; Kingdon, 1984) to either increase or contain the political conflict (Schattschneider, 1978). Fourth, policy actors target policy venues to influence governmental decision makers, mid fifth, the venues are located in multiple levels mid branches of government (Sabatier, 1988; Sabatier & Jenkins-Smith, 1993). Sixth, competing advocacy coalitions use scientific mid technical information to bolster their own arguments and/or to debunk their opponents (Jenkins-Smith, 1988; Sabatier & Jenkins-Smith, 1993). Seventh, contentious policy debates related to a single topic often last for a decade or more mid so the individual policy actors and policy venues involved in the debates are not constant (Sabatier, 1988). Finally, policy brokers may be present and persuade policy actors (both governmental and nongovernmental) to find a policy solution to end the debate (Sabatier, 1988).1
1 The policy broker is one who does not have interest in the outcome of the decision, but does wish for the debate to end and pushes for a decision or negotiation to be made. Another identified role is the policy entrepreneur (Baumgartner & Jones, 1993; Rocherfort & Cobb 1994; Mintrom & Vergani, 1997). The entrepreneur could be considered a policy advocate and part of an advocacy coalition, or a decision maker, or policy broker. The ACF acknowledges there are policy actors who are not part of coalitions, but do contribute to the policy debates within a subsystem (i.e., by making decisions within a policy venue). The policy broker and entrepreneur have similar traits in that they may or may not be part of a coalition and have a vested interest in a policy change occurring. Mintrom and Verganis (1997) argument for differentiation is the entrepreneur actively seeks broad change within a domain or subsystem while a broker or activist seeks specific policy change. Maloney et al., & Olsson (2011) both discuss the insider activist, which is a type of policy advocate.
2


Research on the policy process has taught us a great deal about policy actor beliefs and behaviors. For example, the Institutional Analysis and Development Framework examines how rules and norms influences individual and group behaviors (Ostrom, 2005). The Multiple Streams Framework (Kingdon, 1984), Punctuated Equilibrium Theory (PET) (Baumgartner & Jones, 1993), mid the agenda setting work of Schattschneider (1975) and Rocherfort mid Cobb (1994) each provide valuable insights to how the policy actors advocating for policy change (or stasis) employ different strategies to exploit political opportunities and influence decision makers mid the broader political conflict. Policy actors may strategically select a venue or change venues to expand political conflict, or policy actors attempt to keep discussions at the current venue to contain the conflict (Baumgartner & Jones, 1993; Schattschneider, 1975; Sabatier & Jenkins-Smith, 1993; Schattschneider, 1975; Pralle, 2003). Indeed, interest groups will strategically seek out venues where they think they have the best chances of achieving their policy goals (e.g., Baumgartner & Jones, 1993; Pralle, 2003; Constantelos, 2010). The Advocacy Coalition Framework examines how an individuals beliefs affect their policy goals, provide motivation for action, and act as a heuristic to filter information and identify allies (e.g., Sabatier, 1988).
Each of these theories mid frameworks have strengths mid limitations in their ability to explain policy actor behaviors or policy process outcomes. It is not the goal of this dissertation to highlight or address each of the gaps. However, the four proceeding chapters examine specific gaps in our understanding of policy actor belief or behavior in contentious policy debates. Each chapter of this research uses the Advocacy Coalition Framework (ACF), a prominent theory for describing and explaining policy processes in contentious settings, as its theoretical foundation. The ACF lays out a general logic for policy actor behavior within
3


the policy process. The ACF also provides researchers with clem" conceptualizations of the research setting and the policy actors involved. Finally, researchers can apply multiple theories within the ACF. This research applies theories compatible with the ACF, such as the theory of venue shopping, to develop specific hypotheses when the ACFs logic cannot explain nuances related to each chapters research question or questions.
The remainder of Chapter 1 introduces the contentious policy debate in which each empirical chapter is set: the state level hydraulic fracturing-based oil mid gas development debates in the United States. Then it provides a literature review of the ACF and key concepts from the ACF applied in this dissertation. Last, it introduces the central theme mid question of each independent, empirical chapter.
Research Setting: Statewide Hydraulic Fracturing (Fracking) Debates
One example of a contentious policy debate is the issue of oil and gas development that uses of hydraulic fracturing, aka fracking. Technological advances in fracking and horizontal drilling that began in the 1980s have enabled the economic extraction of oil and gas trapped in porous rock substrates (e.g., tight sands or shale). In the United States, the oil and gas industry used these technologies to expand its operations in the mid-2000s into known shale mid tight sand formation as well as in previously unidentified deposits. By the late 2000s, the industry was experiencing a modern-day oil and gas boom. Fracking-based oil and gas operations expanded in areas accustomed to the industry and into areas unfamiliar with development. Areas unfamiliar with oil and gas development included population centers, residences, and schools. Further, industry found more opportunities to increase operations in environmentally sensitive areas such as lakes, streams, state and national forests, and wildlife preserves. Because of the risks associated with fracking, and where it is
4


being applied, multiple environmental and health groups mid communities mobilized to oppose oil and gas development that sued fracking across the United States. However, due to
the economic benefits, mineral owners, developers, and oil and gas industry groups also
2
mobilized to support the industry.
By 2007 and 2008, states regulators, such as the Railroad Commission in Texas and the Colorado Oil and Gas Conservation Commission, had updated their oil and gas development policies to incorporate the fracking process (Hydraulic Fracturing Information, 2012). In New York, in 2008, the Governor Paterson placed a moratorium on fracking until the New York Department of Environmental Conservation could update its supplemental environmental impact statement to reflect the industrys advancements (Brown, 2011; NYDEC, 2011). In addition, local governments in Colorado, Texas, and New York had also engaged in fracking policy debates. As a result, some local governments made policies that promoted development, and other local governments made policies in opposition to development (Gallaher, 2015; Fracktracker website). Throughout these debates, policy actors fought to get policies changed in their favor (Heikkila et al., 2014). Opponents of oil and gas development argued to stop or limit fracking because the processes negatively impacts the environment and public health mid safety (Food & Water Watch, 2015; Gallaher et al., 2014; Heikkila et al., 2014b; Pierce et al., 2013). Proponents of development downplayed the environmental and health concerns, mid argued to continue fracking because 2 3
2 While several risks related to hydraulic-fracturing based development were not part of the actual process of hydraulic fracturing (e.g., the truck traffic used to bring water to well sites, or methane emissions from well heads), the term fracking became the colloquial phrase used by both sides of the debates to introduce and debate the issue. Fracking is therefore the blanket word used in this work as shorthand for hydraulic fracturing-based oil and gas development.
3 Federal level activity is also present in the United States. For example, debates related to hydraulic fracturing-related regulation are found in both congressional and regulatory venues (e.g., the Bureau of Land Management and the Environmental Protection Agency). Federal activity is not included in this research because the dataset used suggests that federal actors and issues have a played a minor role in state-level debates (as of 2013) (Gallaher et al., 2014; Heikkila et al., 2014b; Pierce et al., 2013).
5


development provided significant economic and national security benefits (In the Matter of Changes to the Rules of the Oil & Gas Conservation Commission of the State of Colorado to Consider Hydraulic Fracturing Disclosure Rules, 2011; Hassett & Mathur, 2013; Heikkila et al., 2014b; COGA, 2014).
The political conflict mid policy debates surrounding hydraulic fracturing-based oil and gas development are like other contentious policy debates in many ways. First, two opposing coalitions are attempting to sway policy outcomes and these coalitions are made of broad range of governmental and non-governmental actors (Heikkila et al., 2014). Representatives from each level of government, alongside environmental and industry interest groups, royalty owners, agricultural representatives, mid concerned citizen groups are mobilizing mid participating in policy discussions. Second, the policy actors on both sides of the debates use scientific mid technical information to support their arguments (e.g.,
Colorado rules and Texas Chemical Disclosure rule making process, and New Yorks process updating their Supplemental Environmental Impact Statement beginning in 2009 through 2014). Third, the policy debates have occurred over long periods. For example, the states where the two technologies were first employed have addressed hydraulic fracturing-based development concerns with policy change since the early 2000s. For instance, Garfield County, Colorado created their Energy Advisory Hom'd in 2004 made of industry, environmental, public, municipal, and county representatives, then in 2005 Garfield County signed a Memorandum of Understanding with the Colorado Oil and Gas Conservation Commission, industry, and environmental representatives in 2005 to evaluate water quality and potential impacts from drilling. Similarly, in 2003 the town of Flower Mound, Texas passed ordinances to regulate distance of wells to other buildings, noise, safety, and
6


environmental impacts, which have undergone multiple major policy change processes since 2007. Additionally, both Texas and Colorado underwent major rule making processes beginning in 2007 mid New York began updating its supplemental general environmental impact statement (SGESI) for oil and gas operations. These states also are re-kindling policy debates from over 30 years ago (Gallaher 2014; Minor, 2014). Finally, the policy debates around hydraulic fracturing-based development resemble other contentious debates in that the policy actors involved are engaging in a wide variety of activities to sway both the publics mid decision makers opinions to achieve their policy change strategies. For example, policy advocates are using media, holding protests, re-defining the issue to gain broader support, mobilizing support from their association members, and engaging multiple policy making venues (Gallaher et al., 2014; Heikkila et al., 2014a; Heikkila et ah, 2014b; Pierce et al., 2013; Gottlieb, 2012; Meyer, 2012; Brush, 2013).
Previous research on state-level fracking policy debates shows local government representatives (e.g., officials from county mid municipal governments) as part of the advocacy coalitions involved in state-level politics (Heikkila et al., 2014a). The advocacy coalitions identified in the statewide policy subsystem are engaging in numerous activities at multiple policy venues at the state mid local levels. Other research on hydraulic fracturing politics highlights local governmental decision makers are engaged in within their own jurisdiction on policy issues related to road maintenance, land use, setbacks, mid environment (Riley, 2007; Groundwater Protection Council & AFF Consulting, 2009). Given that the issue of fracking involves statewide contentious policy debates, and a variety of activity of interest groups and governmental representatives, it is an appropriate policy context to
7


explore policy actor beliefs and behavior. This dissertation examines fracking policy debates within Colorado, Texas, mid New York.
Literature Review: The Advocacy Coalition Framework
The Advocacy Coalition Framework (ACF) is built to describe and explain the actions of policy actors in contentious political contexts. The frameworks focus is on how policy actors form advocacy coalitions to engage in the policy process mid interact with governing sovereigns (i.e., decision makers in policy making venues). For example, The ACF is theoretically geared to answer questions in three areas: questions related to policy change such as what influences policy change and when is policy change likely, questions related to coalition behavior such as why coalition members make particular choices, and questions related to policy-oriented learning such as when does learning occur or how does learning change beliefs. This section describes parts of the ACF relevant to the main concepts used in this dissertation to examine beliefs and behavior of policy actors. Those concepts include the policy subsystem, advocacy coalitions, policy actors, and policy actor beliefs. Policy actor beliefs are also used in this dissertation to compare groups of policy actors within a subsystem. In addition to beliefs, this section introduces three other concepts used in this dissertation distinguish local governmental policy actors from policy actors associated with interest groups. Those include resources for political advocacy, political activities, and networks.
While the ACF has many strengths, mid provides researchers with concepts mid relationships between the concepts to explain the policy process, the ACF also has limitations. For example, it lacks some explanatory power on how advocacy coalitions choose among the available governing sovereigns within a subsystem to deploy their
8


advocacy activities. Additionally, even though the ACF is built upon a model of the individual, it does not engage in explaining or comparing individual level attributes (beliefs withstanding) such as resources, political activities, or networks. When these limitations arise in each individual chapter, this dissertation borrows ideas from other compatible theories to develop its hypotheses and expectations. Specific ACF limitations are identified as each chapter is introduced below.
Policy Subsystems and Advocacy Coalitions
Sabatier (1988), the founder of the ACF, recognized that the study of the policy process and policy change needed to expand beyond a focus on single policy events and decision makers at a single venue. To do so, he proposed that scholars change their unit of analysis to the policy subsystem. He argued that viewing policy processes a policy subsystem enables a broader view on policy processes and change in three ways. First, a substantive policy topic, rather than an event or single decision, is the focal point of a subsystem.
Second, the policy subsystem includes all policy activity within a geographic region and can include multiple political jurisdictions. Third, the subsystem view acknowledges all policy actors engaged in policy processes related to a specific topic (e.g., scientists, interest groups, news media, mid decision makers at all levels of government) mid all policy making venues within geographic boundary (Sabatier, 1988). While the analyst could vary the geographic scope of the subsystem to change the level of granularity of analysis of the policy processes within, typical ACF studies examine subsystems bound by national, regional, or state boundaries. Therefore, once the analyst defines the policy subsystem they can then define internal versus external influences on the policy subsystem, who is and is not a policy
9


actor, and identify the available political venues through which political conflicts may play
out.
The ACF assumes subsystems are largely independent of each other due to the time and resources required for a policy actor to specialize on a topic to engage in a single subsystem (Heclo, 1978). However, outputs from one subsystem can impact other subsystems. In these instances, the outputs from one subsystem are treated as external events on another subsystem. The influence of one subsystems outputs on another subsystem depends on the topic salience and proximity of the two subsystems (Zafonte & Sabatier, 1998). For example, the influence of one subsystems outputs on another subsystem may be stronger when the policy issue of the two subsystems is similar than when subsystems share the same geographic boundary, but focus on different policy issues (Nohrstedt & Weible, 2010). For example, outputs from the Colorado water policy subsystem may inform policy debates in the New Hampshire water policy subsystem, but outputs from the Colorado water policy subsystem would likely have little effect on the Colorado child welfare subsystem. Additionally, policy actors within different subsystems may interact in the search for allies or new policy making venues (Zafonte & Sabatier, 1998). In their research, Lubell, Henry, and McCoy (2010) argue actions like venue shopping increase the interconnections between dissimilar policy issues because a single policy making venue may hold authority over both topics. Therefore, the decisions made on one policy topic could affect decisions made within the same venue on a separate policy topic. The interactions between different policy subsystems are not a focus of this research.
Advocacy coalitions operate within policy subsystems. The advocacy coalition simplifies a researchers analyses of political activity within a subsystem because the
10


coalition conceptually recognizes the multitude of policy actors involved in policy processes (e.g., individuals from all levels of government, interest groups, the scientific community, and the media), but does not require every policy actor to be identified or examined. In other words, when researchers collect individual-level data they can treat the data as representative points and aggregate the information to describe beliefs mid political behaviors of coalitions.
The ACF uses a model of the boundedly rational individual and insights of group behavior from policy network theory to inform the advocacy coalition concept (Sabatier, 1988; Sabatier & Weible, 2007). The ACFs model of the individual assumes information gathering and decisions of the individual are filtered through their personal beliefs and that these beliefs drive their policy preferences. Furthermore, the ACF argues that to overcome individual physical mid cognitive limits, policy actors form coalitions to share resources mid coordinate political activities. Policy actors coalesce into coalitions, in part, by identifying with whom they share similar policy preferences (Sabatier, 1988; Zafonte & Sabatier 1998; Sabatier & Weible, 2007). An advocacy coalition is therefore a broad network of policy actors from local, state and federal governments, interest groups, the scientific community, and the media. These policy actors have individual, belief-driven goals, but choose to act collectively to increase their ability to influence decision makers, mid translate their beliefs into policies.
The nature of interactions between advocacy coalitions within a subsystem ranges from cooperative to conflicting (Weible, 2008). In policy subsystems with a contentious substantive topic there are typically two or three conflicting coalitions (Weible, Sabatier, & McQueen, 2009), but can range between one and five (Weible, Sabatier, & McQueen, 2009). In contentious subsystems, the coalition that maintains political control over policy decisions
11


over extended periods is considered a dominant coalition and acts to keep the status quo or supports policy changes that are congruent with their beliefs and support their goals. When there is a dominant coalition, the opposition, mobilized in one or more minority coalitions, seeks policy change to affect policy in ways that are congruent with their beliefs (Sabatier & Jenkins-Smith 1993; Nohrstedt, 2010). Minority coalitions often seek allies from outside the subsystem or take policy debates to venues that differ from where debates are traditionally held in the subsystem (Fritschler, 1983; Baumgartner & Jones, 1993; Browne, 1990; Worsham, 1997).
Policy Actors
Policy actors within a subsystem are individuals, usually professionally affiliated with an organization, involved in the policy area and dedicating at least some time to influencing either directly or indirectly the politics of the subsystem. In contrast, an individual who submits an official comment on a policy debate, participates in a protest, or votes on a law related to a policy topic is not necessarily a policy actor. In the ACF, policy actors differ from other citizens by the time they devote to an issue and the extent they specialize in the issue. A policy actor may play multiple roles within a subsystem: they can be a policy advocate mid member of an advocacy coalition, they may be a decision maker within a governing body, or they may be a policy broker. There is no reason one policy actor may not take on multiple roles within the same subsystem at different times or in different situations. For example, in one situation a governor may act as a policy broker for a regulatory body attempting to develop policies where competing coalitions are deadlocked on details within the policy. In a second situation, the governor may be the decision maker being lobbied by the competing coalitions.
12


Policy Actor Comparison Beliefs, Resources, Activities, and Networks
Scholars compare advocacy coalitions using four key attributes from the ACF: beliefs, resources, political activities, mid networks. While the ACFs focus is on similarities and differences of coalitions in a subsystem, rather than individual policy actors, those attributes can be applied to examine policy actors and other groupings of policy actors. I argue that beliefs, resources, political activities, and networks can also be used to describe an individuals or an organizations capacity to engage in the policy process, just as they can be used to describe an advocacy coalitions capacity to engage in the policy process. This section briefly introduces the four attributes, leaving the deeper discussions of how the attributes may differ across groups to the relevant empirical chapter.
The ACFs Belief System. Policy actor beliefs are a key attribute of individual policy actors, and a foundational element of the ACF. The ACF views a policy actors beliefs as their motivation for acting within the subsystem and the basis for their policy preferences.
The ACF categorizes an individuals beliefs using a three-tiered hierarchical belief system.
At the highest level of the hierarchy are deep core beliefs, then policy core, and finally secondary beliefs. In this hierarchy, beliefs range from the abstract to the specific (Sabatier & Jenkins-Smith, 1993; Peffley & Hurwitz, 1985).
The researcher can identify a beliefs place on the hierarchy by considering multiple attributes including the beliefs level of abstraction, how empirically based the belief is, the geographic scope of the belief with respect to the subsystem, mid the level of difficulty to change the belief (Sabatier, 1988; Sabatier & Weible, 2007; Weible, Sabatier, & McQueen, 2009). For example, an individuals deep core beliefs are analogous to basic world views mid values. Deep core beliefs are considered constant mid are not related to specific policy topics.
13


Policy core beliefs are thought to be subsystem-wide mid define priorities such as whose welfare matters most in the subsystem, the role of government (including which level of government should regulate), problem identification and its seriousness at the subsystem level, and preferred policy solutions (Sabatier & Weible, 2007; Jenkins-Smith, Nohrstedt, Weible, & Sabatier, 2014). Policy core beliefs are considered difficult to change but may shift over long periods of time of a decade or more (Sabatier, 1998). The lowest level beliefs are secondary beliefs. Secondary beliefs are not subsystem-wide and typically associated with preference for a policy tool or seriousness and cause of a problem in a specific locale (Sabatier & Weible, 2007 p. 196). Secondary beliefs are the most malleable of the three belief types, yet still resistant to change, mid are more easily measured (Weible & Sabatier, 2006).
Overall, individual beliefs guide problem perceptions mid policy preferences, mid are the inception of an individuals policy goals (Sabatier, 1988). Further, an individuals beliefs moderate information processing and act as a cognitive heuristic to identify potential allies (Scholz & Pinney, 1995; Sabatier, 1988).
Beliefs as a Comparative Attribute. Not only are beliefs a defining characteristic of policy actors, but beliefs can be used to describe differences between mid within advocacy coalitions. While the ACFs theory mid empirical evidence from applications of the ACF shows policy actors form coalitions with others who share similar policy core beliefs, there is also evidence that policy actors of the same advocacy coalition have varying policy core beliefs (Sabatier, 1988; Nohrstedt, 2010). ACF scholars attribute coalition-level variation in policy core beliefs, such as policy preferences, to differences in the members individual beliefs (Sabatier, 1988; Weible, 2006; Nohrstedt, 2010) mid to the individuals organizational
14


affiliation (Jenkins-Smith & Claire, 1993; Nohrstedt, 2005; Nohrstedt, 2010). There are endogeneity issues with the organization affiliation argument (Sabatier, 1988), however there is evidence that an individual may have their beliefs or self-interests coopted by their organizations goals. These organizational-level goals then influence a policy actors final policy preferences. For example, the policy preferences of governmental actors are influenced by their interest for continued public support, mid that this interest supersedes their policy core beliefs (Nohrstedt, 2005; 2010).
Resources. A second attribute for comparing policy actors is their resources for political activity. The ACF states that resources give coalitions capacity to pi mi and act on different strategies and support their information processing mid learning (Sabatier & Weible, 2007; Howlett, 2009; Elgin & Weible, 2013). For example, resources provide policy actors or coalitions with the capacity to engage decision making venues (Holyoke, Brown & Henig, 2012). Further, when two or more coalitions are engaged in a political debate, they use resources to influence policy outcomes (Jenkins-Smith, 1988). Resource categories include finances, leadership, access to authority, access to scientific and technical information, and mobilizable supporters (Sabatier & Weible, 2007; Weible 2007). Because a coalitions resources are an aggregation of individual organizational resources, the same logic that resources are a key attribute of coalitions and that those resources allow coalitions to act can be applied to individual policy actors.
Political Activities. The ACF connects the resources and beliefs of coalitions to its strategies to influence decisions by governmental authorities (Sabatier & Weible, 2007, Fig 7.1, pg. 191). This dissertation examines specific political activities rather than broad strategies. If strategies are a plan or method for achieving a goal over a short or long period
15


of time, then political activities can be thought of as the discrete actions of political advocates used to implement a political strategy. For example, a coalition may venue shop as a strategy to achieve a goal of policy change (Pralle, 2003), but within a strategy, many activities may ensue such as lobbying, testifying, or making official comments during policy making processes, or mobilizing troops to engage the decision makers of the venue. While describing the activities of a policy actor or coalition may not reveal their strategy, it does highlight the discrete ways in which they are engaging in the policy process within the subsystem.
Networks. One way to describe the nature of activities and relationships between policy actors is through the policy network literature. For example, an advocacy coalition represents a network of individual policy actors who coalesce through shared beliefs and a desire for policy change mid coordinate activities aimed at achieving a policy goal. Advocacy coalitions, however, are but one type of network that may be found within a subsystem. Advocacy coalitions are like ally and coordination networks described in the policy network literature (Salisbury, Heinze, Laumann, & Nelson, 1987; Zafonte & Sabatier, 1998; Weible & Sabatier, 2005). Other networks found in the policy network literature include power, information mid advice, and resource sharing networks (Weible & Sabatier, 2005). Each of these networks share a common theme related to resource control or exchange and has the potential to include policy actors who are identified in separate coalitions. Indeed, applications of the ACF that examine the interactions of disparate groups acknowledge that policy actors who are members of conflicting advocacy coalitions may interact outside of the policy debates. These policy actors interact to acquire resources or because of institutional or functional links that dictate their interactions (Zafonte & Sabatier, 1998). While the ACF focusses on coordination of individuals as they relate to forming advocacy coalitions, other
16


coordination patterns exist among policy actors within the subsystem. This dissertation does not define the other coordination patterns, but describes the networks of individuals in two ways: first it describes an individuals network pattern by who else they collaborate with, and second, the number of other actors with whom the individual collaborates.
Chapter Introduction
The research for this dissertation is broken into four empirical chapters. Each chapter is written as a stand-alone publishable article. Therefore, some information in the theory and background on the issue of fracking is repeated across the chapters. For example, the research in each empirical chapter of this dissertation applies the ACF to examine policy actor engagement in the policy process within state-level fracking debates.
Chapter 2
Chapter 2 of this dissertation examines questions related to ACFs hierarchical belief system in two ways. First, it models the relationship between deep core, policy core, and secondary beliefs to address a limitation within the ACF regarding how an individuals beliefs interact with one another (Jenkins-Smith et al., 2016). Using the fracking debates in Texas and Colorado as the backdrop, the chapter models how an individuals general attitude toward the role of government in daily life (deep core) mid their normative policy preference related to fracking (policy core) affects their preference for which level of government should regulate issues related to fracking (secondary). Policy process scholars agree that the higher-level, normative, deep core beliefs inform policy core mid secondary beliefs (Jenkins -Smith & Sabatier, 1994; Sabatier, 1998; Sabatier & Weible, 2007; Jenkins-Smith, Nohrstedt, Weible, & Sabatier, 2014). For example, an individuals deep core belief related to the relationship of humans and the natural environment will inform their policy preference on
17


climate change, and potentially their secondary belief on addressing water use in their community. But, the degree to which deep core beliefs constrain policy core mid secondary beliefs remains unclear (Jenkins-Smith & Sabatier, 1994; Weible, Sabatier, & Lubell, 2004).4 Following the same example, ACF theory and empirical evidence from models of the belief system do not describe if an individuals deep core belief that nature is something to protect will constrain their preference for a policy tool to manage local water use (i.e. a secondary belief). Chapter 2 seeks to address this limitation within the ACF regarding how an individuals different beliefs interact with one another by
Second, Chapter 2 examines how the context of the policy debate impacts policy preferences. Specifically, it uses differences in Colorados and Texas regulations estimate how current policies affect a policy actors preference for which level of government should regulate an issue. The ACF acknowledges that the context surrounding the policy issue, such as the nature of the good, the current rules in place, mid physical attributes related to the problem, also constrains policy actors goals mid action (Sabatier, 1988; Jenkins-Smith & Sabatier, 1994). For example, Jenkins-Smith & Sabatier (1994) describe how air policy is affected by the fact that air quality is a collective good mid the physical properties of the earth that impact air flow (pg. 180). Indeed, the logic of the ACF highlights that advocacy groups actions mid goals are strategic and a result of contextual issues surrounding mi issue and the beliefs of policy actors within the advocacy coalitions (Sabatier & Weible, 2007; Jenkins-Smith, Nohrstedt, Weible, & Sabatier, 2014). The theory of venue shopping is applied alongside the ACF to develop hypotheses related to how normative beliefs mid the context of a policy debate affect specific policy preferences.
4 Researchers find it is operationally difficult to differentiate between policy core and secondary beliefs (e.g., Olson, Olson, & Grawronski, 1999).
18


Chapter 3
Chapter 3 examines a strategic behavior of policy actors venue shopping in the fracking debates in New York. Venue shopping is the act of a policy actor engaging a policy making venue (e.g., a state legislature, or court, or a city council) to achieve their policy goals. The current venue shopping literature limits our understanding of policy actors strategic behaviors in two ways. First, the current studies do not include the range of policy actors in a subsystem; rather they typically focus on interest groups (e.g., Holyoke et al., 2012; Ley, 2016; Buffardi et al., 2015; Beyers & Kerreman, 2012; Constantelos, 2010).5 Second, the vertical or multi-level venue shopping research designs are complex and include venues in multiple states or across national boundaries, which draw theoretical focus away from policy actor behavior mid onto institutional effects (e.g., Beyers & Kerreman, 2010; Constantelos, 2010).
To address these limitations in the venue shopping literature, Chapter 3 applies the
Advocacy Coalition Framework (ACF) to examine how a policy actors beliefs affect their
venue shopping behavior. The ACFs theoretical and analytical tools address the two
limitations described above. First, the ACFs definition of the advocacy coalition includes a
broad definition of policy actors involved in the policy process. Similarly, the ACFs
definition of the subsystem includes all potential venues within the subsystem boundaries.
Second, the policy subsystem allows the researcher to place boundaries on the research to
avoid institutional features that could impact venue choices. The state-level subsystem holds
constant those institutional features commonly hypothesized in the venue shopping literature
5 Holyoke et al., (2012), for example, surveyed charter schools in three states boards. Another example is Leys (2016) case study which focused on industry groups in Oregon. Others have examined a set of interest groups within a subsystem. For example, Buffardi et al.s (2015) examination of nonprofits in Seattle, Beyers and Kerremans (2012) study of NGOs, business organization, and labor associations in the European Union, or Constantelos (2010) study of trade, business, or professional associations in Ontario, CA and Michigan, USA.
19


as influential on multi-venue choice (e.g., Constantelos, 2010; Holyoke, Brown, & Henig, 2012; Beyers & Kerremans, 2012), allowing for an empirical focus on policy actor attributes. For example, in this chapter, the ACFs subsystem boundary is drawn around the state of New York. This boundary contains the venue shopping inquiry within the state, mid avoids interstate or international differences in lobbying rules or other rules that may affect how often or why a policy actor shops a venue.
Contemporary ACF research highlights how a coalitions ability to select the right venue can have major impacts in policy change (Nohrstedt, 2011), but the ACF neither describes nor explains the advocacy coalitions selection process between one governing sovereign mid another. The ACFs guidance in venue shopping is limited to the expectation that advocacy coalitions strategically engage governing sovereigns to ensure their decisions align with the coalitions beliefs mid advocacy coalitions require resources to engage in the policy process (Sabatier, 1988). At the individual level, the ACF states decisions and information are shaped by their beliefs mid perceptions. Therefore, Chapter 3 uses the ACF as a starting point and then applies compatible theories within the ACF to develop hypotheses to explain the policy actors choice to engage with governmental venues. As such, venue shopping-related ideas from the work of Schattschneider (1975), punctuated equilibrium theory (Baumgartner & Jones, 1993) mid other related scholarship (Pralle, 2004; Holyoke, Brown, & Henig, 2012; Constantelos, 2010) are applied to build hypotheses on the venue choices of policy actors.
Chapter 3 focuses its inquiry into venue shopping through two common venue shopping question. It asks what factors influence the total number of venues shopped by a policy actor and what factors affect a policy actors shopping frequency at specific venues?
20


Chapter 4 mid 5
Chapter 4 mid Chapter 5 examine how policy actors associated with local governments compare to policy actors associated with interest groups within a contentious policy debate. Policy process research on regional, state, and national environmental policy issues show local governmental representatives are active in policy advocates alongside other stakeholders mid interest groups working to influence policy change (e.g., Sabatier, 1988; Sabatier & Jenkins-Smith, 1994; Weible, 2006; Koontz et al., 2004; Blomquist, Schlager & Heikkila, 2004; Scholz & Stiftel, 2005). However, little is known about how local governmental representatives compare to other advocates within these broad policy debates. The ACF offers some insight into variation among coalition members who are affiliated with government versus interest groups, but the frameworks focus largely remains at the coalition level of analysis. Insights from the ACF regarding coalition beliefs, resources, political activities, and networks are applied to develop expectations related to how policy actors associated with local governments compare to policy actors associated with interest groups.
Chapter 4 asks how the beliefs of local governmental policy actors compare to policy actors associated with interest groups. Chapter 4 builds on previous research that finds policy preferences of individuals are mediated by their organizational affiliation and on one of the ACFs original hypotheses, which states: [w]ithin a coalition, administrative agencies will usually advocate more moderate positions than their interest-group allies (Sabatier, 1998, p. 106). Chapter 5 compares the resources, political activities, mid collaborative networks of local governmental policy actors to policy actors associated with interest groups.
To build the expectation that policy actors from different organizational types will have different resources Chapter 5 uses ideas from the resource based view of the firm
21


(Penrose, 1959; Wenderfelt, 1984) and resource dependence theory (Pfeffer & Salancik, 1978). To compare the political activities of policy actors affiliated with local governments with the resources of policy actors affiliated with interest group allies, Chapter 5 uses ideas from the public management literature to set the expectation that governmental actors will engage in different activities than non-governmental actors (Rainey & Bozeman, 2000; Rosenbloom, 2015). Finally, to compare the networks of policy actors affiliated with local governments with the resources of policy actors affiliated with interest group allies, Chapter 5 examines two aspects of policy actor networks. First, it explores the size of the networks. No expectations are developed for how the size of the network will compare across policy actors associated with local governments and interest groups. Second, it examines the network pattern of policy actors. Resource dependence theory is used to build the expectation that the network pattern of policy actors associated with local governments will be similar to each other and different than the network patterns of policy actors associated with interest groups.
Wrap up
The four independent, empirical chapters are presented next. The final chapter, Chapter 6, provides an over-arching conclusion and discussion of the future direction of my research.
22


CHAPTER II
TRANSLATING BELIEFS: HOW POLICY ACTORS DEEP AND POLICY CORE BELIEFS AFFECT THEIR SECONDARY BELIEFS Introduction: Who Regulates
A policy debate is a situation where individuals and/or organizations engage with governments to change or maintain the governments acknowledgement of and involvement in a problem. The result of a policy debate could include a redistribution of resources, new regulations, or other governmental tools to address a problem. The individuals or organizations involved in the debates use their limited resources to develop problem definitions, policy solutions, form coalitions, and engage with governing sovereigns. In contentious policy debates, there are typically two opposing groups who compete to achieve disparate goals. Policy process scholars examine these policy debates, mid the choices made by those involved, to develop theories that explain individual and group behavior, and policy change. The Advocacy Coalition Framework (ACF) is one of the prominent policy process theory that focusses on group behavior within contentious policy debates. Through the ACF, individual mid group behavior is explained, in part, through their beliefs. The ACF argues that an individuals beliefs provide them motivation to act and that their policy goals reflect their beliefs. While policy scholars have learned much about beliefs, a kind of belief that has not received much attention in policy process research is the preference over which level of government should regulate a policy issue.
Who regulates is a contentious question due to the nature of regulations. Regulations are a specific type of policy tool used by governments to i) control the entry of a firm, its price setting decision, or its production levels, or ii) limit the impact of an economic
23


activity on social or environmental welfare (Salamon, 2002, pg. 119). Regulations are solutions to problems identified as market failures (Weimer & Vining, 2011). Further, regulations are a coercive type of policy tool that governments use to exert their power to restrict a firms actions to produce or secure goods (Salamon, 2002). Regulations not only influence the distribution of resources and actions of individuals or organizations, but also influence local, state, and national economies (Teske, 2004). Therefore, scholars consider regulations a critical area of public policy in the United States (Teske, 2004, p. 5). The decision over who regulates (i.e., which level of government regulates an issue) can be as contested and debated as what is regulated. In the United States, interest groups have fought long and fierce political battles to change which level of government has regulatory authority over atopic such as transportation, commerce, and social welfare (Teske, 2004).6
Policy scholars identify two potential reasons that policy actors attempt to change which level of government has regulatory authority over an issue. First, the policy actors ideology, or attitude toward government, may affect his or her preference for who regulates. Looking back to the inception of the United States, the founders engaged in ideological debates as they outlined the federal system. Their debates centered on where the federal powers would end mid state powers begin (Middlekauf, 2007). These debates continue in modem day policy debates. Political ideologies seen in the Republican and Democratic parties are centered on the size and role of government. Donahue (1997) argues contemporary shifts in regulatory power were observed as the Republican Party under
6 State-level regulatory authority seemed to dwindle after the Civil war and through the Great Depression as federal regulations over transportation, commerce, and social welfare increased and states economic regulations were preempted by federal action. However, Teske (2004) notes that states maintained some power as they were required to implement many federal programs, and then, in the last part of the 20th century, devolution and deregulation at the federal level took its course, providing even more power back to state governments and regulators.
24


President Nixon, President Reagan, and the 1994 Republican Congress successfully fought to devolve federal regulatory authority to the states. However, ideology alone is not likely to completely explain preferences for federal or state level regulation.
A second, a policy actor may desire to shift regulatory authority from one level of government to another for strategic reasons, based on the context of the debate. For example, as federal actors moved to shrink the federal government in the 1990s other interest groups reacted (Donahue 1997; Teske, 2004). Organized labor, environmental, and consumer groups moved toward open venues at the state level to pass policies to shore up changes or fill perceived voids in regulation at the federal level (Teske, 2004). Interest groups may also push for particular regulations at one level of government to preempt regulation from another (Teske, 2004; Hundley, 1986). Finally, policy actors may strategically seek open venues at multiple levels of government to push their policy agendas (Teske, 2004).
This chapter tests how policy actor beliefs translate into a preference for a specific level of government to regulate an issue (referred to as preferred level of government). The Advocacy Coalition Framework (ACF) is used as the theoretical foundation to address this question. The ACFs model of the individual and its hierarchical belief system is applied to develop hypothesis on how beliefs affect their preferred level of government. The theory of venue shopping is also applied in this research to develop hypotheses for how context affects a policy actors preferred level of government. This research examines a policy actors preferred level of government over a range of specific policy issues related to the topic of hydraulic-fracturing-based oil and gas development, aka fracking. The research includes two localized issues mid two broad issues (explained below). The variation in issue breadth is included because if an individuals beliefs can be distinguished based on the level of
25


abstraction, may also be a difference in how individuals conceptualize policy issue, based on the issues level of abstraction.
This research uses empirical data gathered via an electronic a survey of policy actors involved in the state-level fracking debates in Colorado mid Texas. The two state-level fracking debates provide this research a rich testing environment for its question. For example, within the two debates, there are differences between Colorado and Texas in which level of government regulates specific issues. Further, the issues at the center of the fracking debates in both states vary in breadth (e.g., road damage, a more localized issue vs. air pollution, a broader issue). Finally, the topic of who regulates is contentious. Within each state, policy conflicts developed when policy actors attempted to change the level of government that regulates issues related to fracking.
Contributions and Map of the Paper
The hypotheses are tested by building a model of a policy actors preferred level of government using multinomial regression with post estimation marginal effects. The model will estimate the effect of government attitude mid policy preference on preferred level of government for specific fracking-related issue. This research adds to the ACF by empirically testing the relationship of beliefs within its three-tiered hierarchical belief system. Specifically, it tests the relationship of broad, normative beliefs to specific preferences. Second, the research draws in other theoretic ally-backed factors that may affect a policy actors specific preferences. Finally, this research adds to the ACF by applying the model of regulatory preference over a range of localized and broader issues. The variation in issue breadth allows for the factors in the models to be compared mid contrasted mid provides
26


scholars with more information on the nature of context when examining how normative beliefs translate into specific preferences.
The following section builds the theoretical arguments through a review of the ACF and regulator choice. Next, the paper develops its hypotheses. Then, the paper introduces the research setting, the fracking-based oil mid gas development subsystems of Texas and Colorado and explains the similarities and differences in the two states regulatory structure. Next, the paper describes the methodology and results of the analysis. The paper ends with a discussion of the results, evaluation of the hypotheses, and research limitations.
Theoretical Arguments and Hypotheses Development
The Advocacy Coalition Framework (ACF)
The ACFs hierarchical belief system provides this research with a theoretical foundation to explore how broad beliefs may affect more specific preferences, such as a policy actors preferred level of government to regulate an issue. The ACF also acknowledges that contextual factors, such as the nature of the problem and existing rules, affect a policy actors goals and decisions (Sabatier, 1988). Finally, the ACF is compatible with other policy process theories, such as the theory of venue shopping. Therefore, this research can apply insights from the venue shopping literature to examine how contextual factors affect the relationship between of higher-level beliefs and lower level beliefs.
In addition, and with respect to research design, the ACFs construct of the subsystem provides a conceptual boundary around the fracking policy debates. Policy subsystems include multiple policy actors engaging a multiplicity of policy making venues in policy debates over long periods of time. A policy subsystem is defined by a geographic area, a policy topic, mid the actors within the geographical area involved in the substantive topic. In
27


this research, the policy subsystem is defined by state boundaries, the topic of Fracking, and the policy actors involved in fracking-related policy debates. The ACF assumes subsystems are largely independent of each other due to the time mid resources required for a policy actor to specialize on a topic to engage in a single subsystem (Heclo, 1978). With this said, outputs from one subsystem can impact other subsystems, and policy actors in different subsystems may interact (Zafonte & Sabatier, 1998). However, this study, described in more detail below, assumes two independent subsystems for analysis of regulator preference. These are the fracking policy subsystem in Colorado and the fracking policy subsystem in Texas.
Hypothesis 1 Higher Level Beliefs Influence Lower Level Beliefs
The ACF postulates that public policies are made through a process in which boundedly-rational, belief-motivated individuals form advocacy coalitions, and engage governmental decision makers (Sabatier, 1988; Sabatier & Weible, 2007). A boundedly-rational actor has cognitive limits and their beliefs act as a heuristic for processing information (Scholz & Pinney, 1995). Further, an individuals beliefs provide them with
7
criteria by which they measure what is an appropriate policy response to a problem. Therefore, an individuals beliefs are an important part of deriving range of preferences. For example, the preference for what should be regulated, how it should be regulated, which policy making venue to select for a policy debate, mid who should regulate the issue (Sabatier & Weible, 2007).
The ACF categorizes an individuals beliefs using a three-tiered hierarchical belief system. At the highest level of the hierarchy are deep core beliefs, then policy core, and 7
7 The ACF argues beliefs to be such a strong motivator that it allows individuals to overcome collection action problems, but, as Jenkins-Smith, Nohrstedt, Weible, & Sabatier (2014) point out, this premise is underdeveloped in the ACF, and not pursued here.
28


finally secondary beliefs. In this hierarchy, beliefs range from the abstract to the specific (Sabatier & Jenkins-Smith, 1993; Peffley & Hurwitz, 1985). The researcher can identify a beliefs place on the hierarchy by considering multiple attributes including the beliefs level of abstraction, how empirically based the belief is, the geographic scope of the belief with respect to the subsystem, mid the level of difficulty to change the belief (Sabatier, 1988; Sabatier & Weible, 2007; Weible, Sabatier, & McQueen, 2009). For example, an individuals deep core beliefs are analogous to basic world views mid values. Deep core beliefs are considered constant and are not related to specific policy topics. Policy core beliefs are thought to be subsystem-wide and define priorities such as whose welfare matters most in the subsystem, the role of government (including which level of government should regulate), problem identification and its seriousness at the subsystem level, mid preferred policy solutions (Sabatier & Weible, 2007; Jenkins-Smith, Nohrstedt, Weible, & Sabatier, 2014). Policy core beliefs are considered difficult to change but may shift over long periods of time of a decade or more (Sabatier, 1998). The lowest level beliefs are secondary beliefs. Secondary beliefs are not subsystem-wide and typically associated with preference for a policy tool or seriousness and cause of a problem in a specific locale (Sabatier & Weible, 2007 p. 196). Secondary beliefs are the most malleable of the three belief types, yet still resistant to change, mid are more easily measured (Weible & Sabatier, 2006).
Policy process scholars agree that the higher-level, normative, deep core beliefs inform policy core and secondary beliefs (Jenkins-Smith & Sabatier, 1994; Sabatier, 1998; Sabatier & Weible, 2007; Jenkins-Smith, Nohrstedt, Weible, & Sabatier, 2014). ). For example, an individuals deep core belief related to the relationship of humans and the natural environment will inform their policy preference on climate change, and potentially
29


their secondary belief on addressing water use in their community. But, the degree to which deep core beliefs constrain policy core mid secondary beliefs remains unclem' (Jenkins-Smith & Sabatier, 1994; Weible, Sabatier, & Lubell, 2004).8 Following the same example, ACF theory and empirical evidence from models of the belief system do not describe if an individuals deep core belief that nature is something to protect will constrain their preference for a policy tool to manage local water use (i.e. a secondary belief).
To test the relationship between deep core mid secondary beliefs this paper examines the effect of a specific deep core belief on a series of secondary beliefs. Recent scholarship finds individuals general attitudes from cultural cognition theory are consistent with the ACFs deep core beliefs (Ripberger, Gupta, Silva, & Jenkins-Smith, 2014; Jenkins-Smith, Silva, Gupta, & Ripberger, 2014). This research uses one of those general attitudes, the attitude toward government involvement in daily life, as the deep core belief (Kahan et al., 2007; Gastil et al., 2016) mid the preference for who should regulate as the secondary belief. Hypothesis 1 is therefore:
HI: Policy actors who believe governments should be involved less in daily life (deep core) will prefer lower levels of government to regulate issues (secondary belief. Policy actors who believe governments should be involved more in daily life will prefer higher levels of government to regulate issues.
Hypothesis 2 Context Mediates Lower Level Beliefs
The ACF acknowledges that the context surrounding the policy issue, such as the nature of the good, the current rules in place, and physical attributes related to the problem,
8 Researchers have also shown it is operationally difficult to differentiate between policy core and secondary beliefs (e.g., Olson, Olson, & Grawronski, 1999).
30


constrain policy actors goals and actions (Sabatier, 1988; Jenkins-Smith & Sabatier, 1994 Sabatier & Weible, 2007; Jenkins-Smith, Nohrstedt, Weible, & Sabatier, 2014). Another policy process theory, the theory of venue shopping, supports the idea that context mediates a policy actors policy decisions and goals. Venue shopping is defined as the strategic act of a policy actor choosing a policy making venue to hold a political debate. The theory of venue shopping postulates that a policy actors decisions are contingent on their relative position within a policy debate with respect to the status quo (Pralle, 2003). For example, if the policy actor is dissatisfied with the status quo, they will engage in tactics to increase conflict (Baumgartner & Jones, 1993; Pralle, 2006; Schattschneider, 1975). One method a policy actor may use to increase the conflict and disrupt the status quo is to move the policy debate from one policy venue to another (Baumgartner & Jones, 1993). Conversely, individuals who desire to maintain the status quo may engage in tactics to contain a policy conflict, such as blocking another group or individuals attempt to move the debate to new venues (Pralle, 2006).
While the theory of venue shopping focusses on strategic actions, it can help inform this researchers inquiry into the policy actors preferred level of government to regulate a specific issue. Indeed, Teske (2004) found that policy advocates pushed for regulators at the state level over regulators at federal level through a context-informed strategic decision. In other cases, advocates desired to preempt federal regulation. In other cases, advocates wanted to fill voids in the governing structure left open by lack of federal regulation (Teske, 2004). If policy goals and strategic actions can be influenced by contextual factors, then secondary beliefs, such as regulator preference, may be influenced as well. Hypothesis 2 of this chapter maintains the ACFs hierarchical belief system mid applies the idea that context matters.
31


Specifically, Hypothesis 2 draws upon the venue shopping literatures finding that policy actors who desire to expand political conflict will attempt to move a debate away from policy venues where the debates are currently held. Conversely, policy actors who support the status quo will attempt to maintain the debates at the current policy venues (Pralle, 2006). Drawing upon the ACF and venue shopping literature produces an original hypothesis. Hypothesis 2 is:
H2: Policy actors whose policy core beliefs do not align with the status quo are more likely to prefer regulators at levels of government that are different than where they are currently administered.
Lastly, this paper expects the preference for which level of government to regulate an issue to change depending on the nature of the issue. The ACF identifies the nature of the problem at hand as a factor that impacts a policy actors opportunity structure to act on their policy preferences (Jenkins-Smith & Sabatier, 1994). Following Jenkins-Smith and Sabatiers (1994) example of air quality policy, the policy choices available to a policy actor may be affected by the type of good in question and characteristics of the physical environment that relate to the policy issue (pg. 180). Therefore, I expect the type of issue in question to mediate a policy actors preference for level of government. For example, issues with broader externalities, such as air quality are expected to be viewed differently than issues with localized externalities, such as road damage. To examine this expectation, the research includes issues with easily identifiable externalities (explained further in the operationalization section).
32


Research Setting
The topic of hydraulic fracturing-based oil mid gas development, aka fracking, became a national issue following an expansion of the oil and gas industry that spurred from their technological advances in hydraulic fracturing mid horizontal drilling. In the mid-2000s the oil mid gas industry in the United States used these two technologies to expand its operations into known shale and tight sand formation as well as in previously unidentified shale deposits. Because of the success of these technologies, and the current price of oil, the industry quickly found themselves in a modern-day oil mid gas boom. The industry expanded into areas accustomed to the industry, largely rural, extraction-based communities, but they also grew into areas unfamiliar with oil and gas operations. For example, the industry began drilling near population centers, schools and sensitive natural environments. While the industry expansion provided economic benefits, individuals across the country became concerned that fracking-related development would negatively impact their health or the environment.
Because of the risks associated with fracking, multiple groups mid communities mobilized to oppose hydraulic fracturing-based development across the United States. Opponents of oil and gas development argued that hydraulic fracturing and the processes surrounding the extraction technique would negatively impact the environment mid public health mid safety (Food & Water Watch, 2015; Gallaher et al., 2014; Heikkila et al., 2014b; Pierce et al., 2013). Proponents argued that not only were the environmental and health concerns unfounded, but the economic mid national security benefits outweighed the risks (Hassett & Mathur, 2013; Heikkila et al., 2014b; COGA, 2014; In the Matter of Changes to the Rules of the Oil & Gas Conservation Commission of the State of Colorado to Consider
33


Hydraulic Fracturing Disclosure Rules, 2011). In response to the industrys technological advances and these debates, local, state, and federal lawmakers and agencies began updating or creating their oil mid gas policies. The policies addressed a wide range of issues. For example, state agencies in Colorado mid Texas updated their regulations in the mid-to-late 2000s to adjust where drilling could occur; to require industry disclose the chemicals used in fracking fluids; the minimum distance between the wellhead and another building or land feature; and the amount of methane permissible into the atmosphere from a well (Galbraith, 2012; Neslin, 2009). Indeed, fracking had become a contested policy issue in multiple states by the late 2000s.
In addition to the variety of fracking-related issues debated by the opposing groups, there were also debates over which level of government should regulate fracking. In the late 2000s and early 2010s, environmental mid citizen interest groups were lobbying for more federal or municipal level regulations over the oil and gas industry. For example, environmental interest groups led multiple campaigns to develop federal level regulations for the disclosure of fracking chemicals and emissions from hydraulically fracked wells (FracAct 2011 and 2013). In 2012, the EPA released the first air standards for fracking wells and oil and gas pollutants that had yet to be regulated at the federal level (EPA, 2012).9 Further, groups opposed to fracking sought to expand local regulatory authority in the late 2000s and early 2010s (Gallaher, 2015; Silverman, 2014). In response to the efforts by anti-fracking groups to increase federal or local regulation over fracking, state regulators mid pro-fracking groups opposed changes to who regulated the industry. A typical response by state officials
9 Federal level activity includes debates in both congressional and regulatory venues (e.g., the Bureau of Land Management and the Environmental Protection Agency). Federal activity is not included in this research because the dataset used suggests that federal actors and issues have a played a minor role in state-level debates (as of 2013) (Gallaher et al., 2014; Heikkila et al., 2014b; Pierce et al., 2013).
34


or industry groups to the local level oil mid gas policy is a lawsuit against the municipality (Gallaher, 2015; Sandberg, 2012; Fehling, 2015). Industry associations and other industry representatives often noted that the state was the preferred level of regulation (COGA, 2015). They argued that the industry was too nuanced to be regulated by the federal government (COGA, 2015), but local governments were not equipped to regulate the industry (Sandberg, 2012). Pro-industry groups argue local level regulations would create an unmanageable patchwork of regulation that would hurt the industry (UOGR, 2014; Haley, n.d.; COGA, 2016).
The fracking-related policy debates provide this research with a setting to examine how an individuals beliefs at different levels of the hierarchy mid the context of the debate interact. Within the debates, individuals represent a range of positions on whether fracking should continue and are concerned about many problems related to fracking and oil and gas development that vary in scope (Weible & Heikkila, 2016; Heikkila et al., 2014b). For example, some individuals are concerned about dust and noise near a well, while others are concerned about fracking chemicals contaminating the water table (Heikkila et al., 2014a). In addition, individuals have varying preferences over which level of government should regulate fracking (Heikkila et al., 2014b).
Finally, the current legal structure around oil and gas development provides this research with a regulatory landscape that is spread over multiple levels of government. For example, in the United States, state governments hold most regulatory authority over oil and gas development. However, within the two states used in this research, local governments have varying authority over nuisance issues (e.g., noise and dust) mid the distance a well can be from neighboring buildings. These nuances are explained next.
35


Two Similar Subsystems: Colorado and Texas
Two state-level policy subsystems, the fracking-related oil mid gas subsystems in Colorado and Texas, are used to examine regulatory preferences of statewide policy actors. Colorado and Texas were selected because of their strong similarities related to oil and gas development and regulation and similarities in the reaction to the rise in development due to new techniques of fracking and horizontal drilling. However, Colorado and Texas have different regulatory structures with respect to local-level issues. In this sense, they are considered most-similar cases (Gerring, 2007) mid appropriate for hypothesis testing the affect current regulation has on a policy actors preferred level of government to regulate an issue.
Regulatory Similarities and Differences
Both states have similar histories with oil and gas development and similar regulatory structures in terms of the state agency developing and promulgating rules for development activities (STRONGER, 2011; STRONGER, 1993). Both Colorado and Texas have oil and gas activity dating back to the 1800s and recent booms attributed to horizontal drilling mid hydraulic fracturing innovations mid discoveries of shale deposits. Texas and Colorados state legislatures give authority to a state-level regulatory body: The Railroad Commission of Texas (RCC) and the Colorado Oil and Gas Conservation Commission (COGCC), respectively. Both states have had multiple rule changes to update their regulations to accommodate the new technologies of hydraulic fracturing mid horizontal drilling. They each have addressed disclosure of fracturing chemicals in 2011 (first in Texas and shortly followed in Colorado), but do address issues independently and with slight variations (i.e., Texas has had more focus on water recycling and seismic activity and Colorado has
36


examined the distance between wells and public or private structures and water monitoring). Both Colorados COGCC and Texas RCC oversee the rule change process and policy actors from industry, environmentalists, local representatives, and other organizations make comments during the rule making process.
Regulation in Texas. The RRC regulates a wide range of activities related to oil mid gas operations. At the time of this study rules mid regulation for oil and gas regulation were contained in the Texas Administrative Code Part 1 Title 16 § 3. The RRC rules contain chapters regarding practice and procedure, informal complaint procedure, oil and gas division, environmental protection, carbon dioxide, gas services division, pipeline safety regulations, LP-gas (liquefied petroleum gas) safety rules, surface mining and reclamation division, coal mining regulations, regulations for compressed natural gas, regulations for liquefied natural gas, alternative fuels research and education division, underground regulations for liquefied natural gas, alternative fuels research and education division, underground pipeline damage prevention, mid administration. Most notable is chapter three, which contains the regulations of the Oil mid Gas Division. Chapter 3 section 1 through 107 cover rules ranging from water protection mid other environmental management (§3.8, §3.22, §3.91, and §3.93), well design; including casing, cementing, drilling, and completion requirements(§3.13), plugging and other completion activities (§3.14- §3.16), testing during and after well drilling (§3.17), safety mid emergency management (§3.20, §3.21, and §3.84), hydraulic fracturing chemical disclosure (§3.29), well spacing and density (§3.29, §3.38), waste (§3.98), fees, taxes and exemptions (§3.50, §3.78, §3.83, §3.101, §3.102, mid §3.103), and penalties (§3.107).
37


The Texas Commission on Environmental Quality is another key regulator of oil and gas operations. The TCEQ is the environmental agency for the State of Texas whose primary goals are to ensure clean air, clean water, and the safe management of waste. The TCEQ mid RRC share responsibility over waste, water quality, and injection wells. Since 1982, the TCEQ and RRC have used memorandums of understanding to clarify duties related oil mid gas. In 2011, the MOU was updated again and passed responsibility for providing surface casing and/or groundwater protection recommendations for oil and gas activities... as well as moving the TCEQs Surface Casings Program mid staff to the RCC and renamed the program as the Groundwater Advisory Unit (Railroad Commission of Texas 16 TAC Chapter 3 Oil and Gas Division, 2012). With respect to waste, mid as of the time of this research, the TCEQ had responsibility over solid waste, which excludes waste resulting from oil and gas exploration, development, and production, and the RRC had jurisdiction over oil and gas waste. With respect to water quality, TCEQ sets water quality standards, and RRC enforces standards related to discharges and storm water resulting from oil and gas activities. TCEQ had jurisdiction over other water issues. For example, any surface water diverted for use in hydraulic fracturing must obtain water rights through the TCEQ. Groundwater rights are obtained through courts and the State Legislature and managed either under the rule of capture, through individual land owners, or by Groundwater Conservation Districts (Hydraulic Fracturing Frequently Asked Questions, http://www.rrc.state.tx.us/about/faqs/hydraulicfracturing.php).
Regulation in Colorado. The Colorado Oil and Gas Conservation Commission is the regulating body in Colorado for all oil and gas and is housed in the Department of Natural
38


Resources.10 011 Initially the mission of the COGCC was to promote the oil mid gas industry mid to prevent waste of oil and gas resources. Over time, the mission was modified to include the protection of public health, safety, and welfare and the environment ( Pasternak, 1999). The Commissions regulatory authority includes the application process mid approval for permitting and drill site selection; the planning, site preparation, extraction, clean-up, and surface recovery processes; environmental, health, and public safety requirements; mid the required (consultations) or reactive (complaints and hearings) communication between the operator, land owner, mineral owner, downstream water users, the Colorado Department of Public Health mid Environment, and the Colorado Division of Wildlife (COGCC Rule 201).
In 2008, the COGCC completed a major overhaul of its oil and gas regulations to accommodate process changes related to mid concerns of hydraulic fracturing mid horizontal drilling. The COGCC rules specific to hydraulic fracturing include: Rule 205 inventory chemicals; Rule 317 Well casing and cementing; Cement bond logs; Rule 317B setbacks mid precautions near surface waters and tributaries that are sources of public drinking water; Rule 341 monitor pressures during stimulation; Rule 608 Special requirements for CBM wells; Rules 903 & 904 pit permitting, lining, monitoring, & secondary containment; and Rule 906 requires Commission, CDPHE mid the landowner of any spill that threatens to impact any water of the state ( Hydraulic Fracturing Information, 2012).
The Colorado Division of Water Resources (DWR) is responsible for surface mid groundwater use and consumption. Operators must maneuver through the water use program
10 In 1951, the Colorado General Assembly enacted the Oil and Gas Conservation Act and created the Colorado
011 and Gas Conservation Commission (COGCC) to carry out the provisions of the Act. The COGGC does not regulate exploration and extraction activity on Indian trust lands and minerals or the Southern Ute Indian tribe within the exterior boundaries of the Southern Indian Reservation.
39


administered by the DWRto lease or purchase water rights. Under certain conditions, the Colorado Division of Wildlife (CDW) must be consulted in the development of plans for multiple or individual well location assessments. Further, the Colorado Department of Public Health and Environment (CDPHE) is involved in the permit-to-drill application process when a Local Government Designee (LGD) requests their participation, when the operator seeks a variance to specific rules related to the protection of public health, safety, welfare, or the environment, or when the operator requests to increase well density (306.d. 1. A.ii and 306.d.l.B). The Water Quality Control Division (WQCD) within the CDPHE is responsible for the permitting of discharges to surface waters (Stronger, 2011). The COGCC has a Memorandum of Agreement with the WQCD that gives the COGCC reporting mid initial oversight responsibilities to field inspectors, which include spill mid discharges associated with hydraulic fracturing (Stronger, 2011).
Differences between Colorado and Texas. While local regulatory issues appear similar, a major difference between the two states is in the role of local governments as regulators for local-level issues. In Texas, state-level regulators do not have jurisdiction over roads, leases, pipeline easements, royalty payments, setback distances between the well and other buildings or natural features, or nuisance issues, such as traffic, noise, odors, (RCC website, n.d.). Rather, municipal governments in Texas have the authority over these issues. As such, many cities have developed or amended their ordinances regarding the exploration and production of oil mid gas to include these issues (Barnett Shale Energy Education Council, n.d.). In Colorado, on the other hand, local authority is limited to areas outside of drilling operations like road use and building permits. Nuisance issues such as noise, dust, and odors, and the issue of setback distances are regulated by Colorados state regulatory
40


body, the COGCC. The variation in local authority provides a setting to test hypotheses with respect to policy actors regulatory preference.
In Colorado, debates over local mid state regulatory control over oil and gas development date back to the 1980s and have led to multiple lawsuits between state mid local governments. Local governments have placed moratoriums on drilling mid attempted to use their land-use and zoning authority to dictate where drilling could take place (Gallaher,
2015). State government representatives and the oil and gas industry argue the state has preemptive rights when it comes to oil and gas development: it is in the states interest and therefore local governments cannot intervene. One arc of Colorados local control debates peaked in 2014 when interest groups for local regulations petitioned for multiple state-level ballot initiatives aimed at changing the states constitution to allow for more local control over the industry (Richardson, 2014; Hostetter, 2014). At this time, Governor John Hickenlooper stepped into the fray to negotiate a compromise. A task force to investigate local control was created, and the ballot initiatives mid current lawsuits were dropped as a result (The State of Colorado, 2014). Texas has had some similar responses to local attempts to increase their regulatory purview over development. However, the moratoriums set by Texas local governments have been short-lived (Flower Mound) and, not until after this research was completed, and a ban on fracking-related oil and gas development was set in Denton, TX. This ban was quickly overturned by a state lawsuit (Baker, 2015). Immediately following the overturned local ban, state officials attempted to cut off any future local action through by proposing a bill to make any attempt at local bans on hydraulic fracturing s illegal (Baker, 2015).
41


Methods
Population and Sampling
A team of researchers, including the author, collected survey data used in this paper, as part of a larger project that encompassed Texas and Colorado. In this effort, we conducted two sequential cross-sectional surveys in Colorado and Texas, in 2012 and 2013, respectively. We targeted policy actors involved in the statewide oil mid gas subsystems in Colorado and Texas. Policy actors, in contrast to the general population, defined as individuals who are professionally affiliated with an organization, involved in the policy area, and dedicate at least some time to influence, either directly or indirectly, the politics of the subsystem (Sabatier, 1988; Baumgartner & Leech, 2001).11 We identified policy actors using a modified snowball sampling method. We began by identifying policy actors through internet searches of government documents, such as participant lists in fracking-related rule making and legislative hearings. Next, we expanded the policy actor list by searching on-line newspaper reports mid documents published the policy actors we previously identified for additional names or organizations. Finally, we interviewed a subset the policy actors mid asked them who should be included in the study. Through this process, we checked for biases in the search method by examining the organizational affiliation of the policy actors we identified. Our goal was to have a range of policy actors from the oil and gas industry, environmental groups, local, state, and federal governments, and the scientific community. We adjusted our search criteria to ensure our final policy actor list represented those different organizational affiliations. These methods reduced the possibility that our population sample
11 An individual who submits an official comment on a policy debate, participates in a protest, or votes on a law related to a policy topic is not necessarily considered a policy actor. In the ACF, policy actors are differentiated from other citizens by the time they devote to an issue and the extent they specialize in the issue. Policy actors are differentiated from one another by attributes such as beliefs and resources and by behaviors.
42


had coverage error, or the omission of key policy actors involved in hydraulic fracturing within each state (Singleton & Straits, 2010). Non-probability sampling, such as this, is appropriate when there is not a pre-made list, or other documentation, from which to sample or to create a sampling frame (Singleton & Straits, 2010).
In total, we identified 398 policy actors in Colorado mid 324 policy actors in Texas. Given the sample population size, we sent each policy actor an online survey (Singleton & Straits, 2010). Each survey was created mid distributed through Qualtircs, an online survey tool. We gave each respondent three reminders to complete the survey after the initial request. We received a survey response from 142 of the 398 policy actors in Colorado (a 35.7% response rate) mid survey responses from 78 of the 324 policy actors in Texas (a 24% survey response rate). The lower response rate in Texas limits the generalizability of the results from this survey.
Variable Operationalization
The next section describes how I operationalized the key concepts used to test this papers two hypotheses from survey data of policy actors in Texas and Colorado.
Secondary belief preferred level of government (Dependent Variable for HI and H2). To measure secondary beliefs, I used a survey question that asked If you were to select only one level of government to regulate the following issues related to shale development, which you would prefer, if any? Response categories include no regulation, local government; state government, or Tederal government. The issues included in this paper are public nuisance issues, setback distances, air emissions monitoring, and water 12
12 Texas survey included municipal government and county government. However, the Colorado survey only included local government. Therefore, I combined the municipal and county government preference from the Texas survey into a single local government category.
43


quality monitoring. Each answer is considered a secondary belief and was evaluated separately in a multi-nominal regression model. A policy actors preference for which level of government should regulate a specific issue is considered a secondary belief because it is narrow in scope with respect to the subsystem, and more easily measured (Sabatier &
Weible, 2007).
I include four issues in this research that have distinguishably different externalities. The issues of water quality and air emissions are considered issues with broader externalities. During the 2011-2012 chemical disclosure rule makings in both Colorado and Texas, policy actors who were against fracking argued that fracking fluids could contaminate the water table if a well-casing failed, or surface waters if a spill occurred. With respect to air emissions, policy actors as early as 2008 in Colorado were concerned that oil and gas operations were decreasing regional air quality through the release of volatile organic compounds (CDPHE, 2012; Dunn, 2013). The issues of setback distances and nuisance issues are considered to have more localized externalities. The nuisance issues were clarified in the survey to mean dust, noise, mid light from the well-site. Nuisance issues logically will only affect individuals who are near a specific well. Similarly, the issue of setbacks is considered a localized issue for the fact that the concern is if an accident occurs at a specific well, the local proximity is at risk for being damaged. While there may be differences in how one interprets the scope of these issues, I present the results of the preference for level of government models by pairs related to how I have categorized the issues externalities: water and air issues are presented together and nuisance and setback issues are presented together.
Deep core belief attitude toward government (Independent Variable for HI).
44


To measure the respondent deep core belief about government, the survey asked two questions from cultural cognition theory about their general attitudes toward government involvement in daily life (Kahan et al., 2007; Gastil et al., 2016). First, the government should put limits on the choices individual can make so they do not get in the way of what is good for society. Second, the government should do more to advance societys goals, even if that means limiting the freedom mid choices of individuals. Respondents provided their answer on a four-point Likert scale: -2 = Strongly Disagree, -1 = Moderately Disagree, 1 = Moderately Agree, 2 = Strongly Agree.
These two questions measure respondents levels of individualism vs collectivism (Kahan et al., 2007). Individualism and collectivism are two extremes on a scale measuring an individuals belief on how much choice a person should have in making decision vs. how much governments should intervene to aid in achieving cooperation (Gastil et al., 2016). This research uses agreement to these two questions to indicate that the respondent believes individuals should have less freedom of choice and governments should have a greater role in influencing individual decisions.
Iterated Principal Component Factor analysis with Varimax rotation was used to combine the two scores and create a score reflecting the respondents general attitude toward governmental intervention. A negative score indicates the respondent believes governments should have less influence over citizens decision making, and a positive score indicates the
13
respondent believes government should have more influence over citizens decision.
Policy Core Belief: Policy change preference (Independent Variable for H2). 13
13 Factor analysis resulted in a single factor, with eigenvalue 1.33 and each variable factor loading with a 0.877 after rotation. The factor ranged from -1.12 to 1.69. See Table 2 in Chapter 2 Appendix.
45


Next, the survey asked about a general policy preference toward the fracking in relationship to the status quo. The survey asked Please indicate what comes closest to your current position in relation to unconventional shale development that uses hydraulic fracturing. It should be.... stopped, limited, continued at the current rate expanded moderately, or expanded extensively. I combined the respondents who stated stopped or limited into a single group stop/limit. I kept respondents who stated that fracking should continue at current rate separate. Finally, I combined respondents who stated fracking should expand moderately or expand extensively into an expand group.
Reflecting the ACFs belief system, I categorize the respondents position on fracking as a policy core belief (Sabatier, 1988, pg. 145, Table 1; Sabatier, 1998; Sabatier & Weible, 2007; Jenkins-Smith, 1994; Jenkins-Smith, Nohrstedt, Weible, & Sabatier, 2014; Weible, Sabatier, & McQueen, 2009). The respondents position on fracking aligns with the definition of policy core beliefs as it is: Subsystem wide in scope; highly salient as it is a fundamental policy position related to the substantive topic; at the level of abstraction of this preference is low enough to be focused on a topic, but remains general in that it is referring to a broad range of activities; and related to a policy that if changed, it would be considered a major policy change (i.e., if the current policies that allow fracking were changed to stop fracking).14
Opponents of oil and gas development viewed both the regulations and state regulators as pro-development. Therefore, the stop or limit group was considered to represent those policy actors who were against the status quo given that the national legal and
14 Note: It is also possible to categorize the respondents position on fracking as a policy core policy preference. Policy core policy preferences are derived from policy core beliefs and are a normative position that projects] an image of how the policy subsystem ought to be (Sabatier & Weible, 2007, pg. 195).
46


regulatory framework allowed fracking mid the practice had been expanding throughout the time of the survey. The continue at current rate group is the status quo group.
The status quo in the fracking debates varies by state because of the variation in how states regulate fracking. Nuances in state regulatory structures are described next.
Control Variables. Two control variables were used: the state and the organization type of the respondent. Designating the subsystem from which respondent came from, either Colorado or Texas, provided this paper with the ability to compare between issues that are currently regulated at the state vs. the local government. At the time of this research, state-level agencies were the primary developers mid administrators of fracking-related oil mid gas development rules and regulations. However, Colorado and Texas differed on two local level issues: setback distances and nuisance issues. In Texas, municipalities controlled regulations over these issues. In Colorado, the state agencies controlled regulations over these issues.
It is possible that other differences between the states could affect the strength of the arguments made using this variable. For example, institutional structures, such as legal standing may impact regulator preference (Jones, 2001). From the venue shopping literature, policy actors may choose the venue that has the legal standing, or legal ability, to make policy (Holyoke, Brown, & Henig, 2012) or the venue that is already active in addressing the issue (Mahoney & Baumgartner, 2009).
Additionally, six organization types are included: environmental groups, oil and gas industry groups, federal government representatives, state government representatives, local government representatives, and other. Other includes groups such as academics and consultants and the media.
47


Model Selection mid Analysis Techniques
A multinomial logit model in STATA is used to test the relative influence of each variable on the regulatory preference of the respondents for specific issues associated with fracking. Marginal effects, a postestimation analysis, is then used to estimate the overall effect of each independent variable, mid interaction effects of independent variables, on regulator preference and test the papers hypotheses.
The multinomial logit equation for a policy actors preferred regulator is as follows:
Policy actors preferred regulator model: Preferred level of government to regulate the issuej = Governmental attitude (HI) + Policy preference (H2) + State (Control) + Organizational type (Control).15
Multinomial regression is an appropriate selection for modeling nonlinear systems with two or more categorical dependent variables (McFadden, 1999). Multinomial analysis is similar to a bivariate model (e.g., a logistic regression) in that the coefficients in the model outputs describe the effect on one outcome of the dependent variable with respect to another outcome. However, because multinomial models have more than two outcome choices, a base outcome option is selected by the user, and the coefficients for each independent variable represents a relative probability of selecting the output option of interest, with respect to the base outcome option. But, the probability of one outcome is also impacted by the change in probability of the other outcomes. Therefore, because of the number of equations involved in a multinomial regression, the interpretation is multi-layered. Because of these the interrelationship of the coefficients mid exponentiated coefficients in
15 STATA code: mlogit issue; c.gov_attitude i.policy_preference i.state 1.org_type, base(3) rrr. Variable names are shortened and If statements to eliminate duplicate cases in the dataset are omitted for simplicity.
48


multinomial regression models, Rodriguez (2017) asserts to reach conclusions about actual probabilities [of coefficients] we need to calculate continuous or discrete marginal effects. Further, Williams (2012) notes the postestimation tool of marginal effects provide information on the practical significance of the findings and moves researchers discussions beyond sign mid statistical significance.
A marginal effects analysis is a postestimation tool for regression models and particularly helpful for interpreting the effects of independent variables in multinomial regression (Williams, 2012; Rodriguez, 2017). Predictive margins, or marginal effects, provides the effect of the independent variable on the categorical outcome mid whether the effect is statistically significant.16 Marginal effects can be used to interpret continuous mid categorical independent variables, but it should be noted that the interpretation of results is more straight forward for categorical independent variables (Williams, 2017a, 2017b; Royston, 2013). For categorical independent variables, the marginal effects analysis describes the discrete change in the predicted probability of a specific outcome, given the IV is = 1 mid other variables are held constant. For continuous independent variables, the marginal effect is the instantaneous rate of change and is not always interpreted as the effect of a one-unit increase of the IV on the DV (Willi mils, 2017a). Therefore, different marginal effects operations are considered in this research based on the nature of the independent variables for each hypothesis. Marginal effects are displayed graphically to provide more intuitive information on the effect of each variable on the probability of a specific outcomes (Jann, 2013).
16 Predicting margins allows the user to fix a variables value and maintain the other covariates in the model, getting a predicted value of the dependent variable, and then averaging the predicted value (Becketti, 2003). This method is preferred for multinomial models as interpreting coefficients for multinomial models is difficult as they are discussed as relative probabilities (Rodriguez, 2017).
49


Analysis and Results
First, this paper presents the general results of the four full models mid then describes the results related to Hypothesis 1 and then Hypothesis 2. Lastly, exploratory analysis with the control variables are presented. The full models, presented below in Figure 3.1 mid Figure 3.2, show the marginal effects (dydx) of each variable on the predicted outcomes of for local, state, or federal regulator preference.17 18 Figure 3.1 shows the marginal effects for the localized issue models: setback distances and nuisance issues. Figure 3.2 shows the marginal effects the broader issue models: for air emissions mid water quality. Each figure shows three columns one for each potential outcome of the dependent variable local government, state government, mid federal government. All models show the marginal effects of each variable with 90% confidence intervals which is a visual indicator for statistical significance at a p-value of 0.90 or better.19 To interpret the categorical variables, use the first variable in each group of categorical variables as the comparison variable. For example, the marginal effect of the policy preference of continue is in comparison to the policy preference of stop/limif. The marginal effect of the categorical variable is the change in probability that the respondent would choose the level of government when compared to the comparison value. The marginal effect of the continuous independent variable, governmental attitude, is the instantaneous rate of change of the independent variable on the dependent variable. (Williams, 2017a).
17 Descriptive statistics of the models variables are presented in Appendix A. Note that no regulation was an option, but only three respondents indicated they desired no regulation for the issue of setbacks, two respondents for the issues of nuisance issues and air emissions and only one respondent for the issue of water quality. Therefore, the no regulation response was not included in the output options.
18 Model used Average Marginal Effects (Williams, 2011).
19 Appendix A Chapter 2 Table 4provides the model results in tabular format with the exponentiated coefficients (relative risk ratios) and the exact p-values.
50


Marginal Effects on Regulator Preference: Setbacks
Local State Federal
Government Attitude-
Policy Preference stop/limit-continue-expand-
State Colorado-Texas
Organization Type Environmental groups-oil and gas federal gov-state gov local gov-other
Marginal Effects on Regulator Preference: Nuisance Issues
Local State Federal
Government Attitude-
Policy Preference stop/limit-continue-expand-
State Colorado-Texas
Organization Type Environmental groups-oil and gas federal gov-state gov local gov other
Note: 90% confidence levels shown. For each category of categorical variables, Policy Preference, State, and Organization Type, the first variable listed is the comparison variable, stop/limit, Colorado, and Environmental groups, respectively. See Appendix for full model tables with relative risk ratios.
Figure 3.1. Marginal effects (dydx) of Multinomial Regression Model: Setbacks and Nuisance Issues.
51


Marginal Effects on Regulator Preference: Air Quality
Government Attitude-
Policy Preference
st op /limit -continue-expand-
State
Colorado-
Texas-
Organization Type Environmental groups-oil and gas federal gov-state gov local gov-other
Local
State
Federal
Marginal Effects on Regulator Preference: Water Quality
Local State Federal
Note: 90% confidence levels shown. For each category ofi categorical variables, Policy Preference, State, and Organization Type, the first variable listed is the comparison variable, stop/limit, Colorado, and Environmental groups, respectively. See Appendix for full model tables with relative risk ratios and exact p-values.
Figure 3.2. Marginal effects (dydx) of Multinomial Regression Model: Air emission and Water quality.
52


Overall Results
Governmental attitude. For the issues of nuisance and setbacks, results show that respondents government attitude has no effect on which level of government they prefer to regulate (Figure 3.1). For the issues of air emissions mid water quality, however, results show that respondents government attitude has a significant effect on the respondents preference for local and federal regulators (Figure 3.2). These differences are examined in-depth in the Hypothesis 1 section.
Policy Preference. For the issues of nuisance and setbacks, the results show that respondents who wish for fracking to continue or to expand are less likely to prefer local regulators and more likely to prefer state regulators, when compared to the stop/limit group
(Figure 3.1). Those who wish for fracking to expand are less likely to prefer federal
20
regulators when compared to the stop/limit group.
For air emissions and water quality issues, the continue and expand groups are less likely to prefer local and federal regulators and more likely to prefer state regulators, when compared to the stop/limit group (Figure 3.2). The differences are more pronounced and statistically significant for the issue of water quality. These differences are examined in-depth in the Hypothesis 2 section.
State. For nuisance and setback issues, respondents form Texas are statistically more likely to prefer local regulators and less likely to prefer state regulators for setbacks than those from Colorado. For air emissions mid water quality respondents form Texas are statistically less likely to prefer state regulators and more likely to prefer federal regulators 20
20 The stop/limit group was used as the baseline group for the overall models. Given that the status quo depends on the state, the comparison between status quo is only provided for H2.
53


than Coloradoans. These relationships are considered further in below in the section on Hypothesis 2.
Organization type. For nuisance and setback issues, Oil mid gas industry respondents are more likely to prefer state level regulators and less likely to prefer local or federal level regulators than the environmental groups. The differences between governmental and other organization types with the base group vary between the issue of setbacks and nuisance issues. Statistically significant differences are primarily found in the preference for local or federal level regulators. For air emissions and water quality, environmental groups generally less likely to prefer state regulators and more likely to prefer federal regulators than the other organization types. The only exception is federal government representatives are more likely to prefer federal regulators than the environmental groups for water quality issues.
54


Hypothesis 1
Recall, Hypothesis 1 is: Policy actors who believe governments should be involved less in daily life (deep core belief) will prefer lower levels of government to regulate issues (secondary belief). Policy actors who believe governments should be involved more in daily life will prefer higher levels of government to regulate issues.
To test Hypothesis 1, this paper uses the average marginal effects of a respondents
21
attitude toward government on their preferred level of government to regulate an issue.
Figure 3 shows the instantaneous rates of change of governmental attitude on the preference for local and federal regulator for air quality, water quality, nuisance issues, and setback distances (Williams, 2017a & 2017b). The marginal effect (dydx) value summarizes how a change in the response is related to a change in the covariate (Williams, 2017a, p. 4). The independent variable of interest, government attitude, is continuous from -1.2 to 1.7. The further the government value is from zero, the greater the effect of the covariate on the predicted outcome. If we assume a linear relationship throughout values of the governmental attitude we can interpret the results as such (Figure 3.3). For the issue of air emissions, the results show that a one-unit increase in governmental attitude is associated with a 0.082 (8 percentage points) increase in the probability of preferring federal regulator. Therefore, for the issue of air emissions, respondent who prefer more governmental intervention are more likely to prefer federal regulators. Further, for the issue of air emissions, a one-unit increase in governmental attitude, is associated with a 0.055 (5.5 percentage points) in a respondents 21 22 23
21 See Appendix A, Chapter 2 Table 1 and Table 4 for more detail on respondents preferred regulator for each issue and the tabular data on instantaneous rate of change of governmental attitude on preference for a specific regulator.
22 ST AT A code: margins, dydx (gov_attitude) pr(out(local_regulator)) and margins, dydx (gov_attitu.de) pr(out(federal_regulator)).
23 See Appendix A, Table 4 for the marginal effects of government attitude at representative values.
55


likelihood to prefer local government regulators. Both effects are statistically significant. For the issue of water quality, the results are similar to air quality. However, the effect of government attitude is only statistically significant on a respondents likelihood to prefer local government regulators.
For the small-scale issues of nuisance issues mid setbacks distances, the results show that the effect of respondents deep core belief attitude toward government on their secondary belief their preference over which level of government should regulate is negligible and not statistically significant.
Overall, these results indicate that the effect of a respondents attitude toward government on regulator preference varies by issue type. The results partially align with the hypothesis that policy actors who believe governments should be involved less in daily life prefer lower levels of government to regulate oil and gas issues.
56


0.100
Broad Externality
Localized Externality
0 Air
x Federal Preference, \ 0.082***
0.050
? 0.000 5*
-0.050
__Water_____
Federal Preference, 0.031
A
Nuisance
Federal Preference,
0.003
Local Preference, -0.012
Setbacks
Federal Preference,
0.007

Local Preference, 0.005
Local Preference, Local Preference, A -0.055*** -0.065***
-0.100
P-value <0.10, ** <0.05, *** <0.01.
Figure 3.3. The instantaneous rate of change of governmental attitude on local and federal regulator preference for issues with broad and localized externalities.
57


Hypothesis 2
Recall, Hypothesis 2 is: Policy actors whose policy core beliefs do not align with the status quo are more likely to prefer regulators at levels of government that are different than where they are currently administered.
To test Hypothesis 2, this paper examines the issues with localized and broad externalities separately. Recall that setbacks and nuisance issues were regulated at the state level of government in Colorado and at the local level of government in Texas, mid air and water quality issuers were regulated at the state in both Colorado mid Texas. Therefore, the status quo is different for issues with localized externalities, by state. For those from Texas, the expectation is respondents who are against the status quo those who desire fracking to be stopped/limited will be less likely to prefer local regulators than respondents who are for the status quo those who desire fracking to continue at its current rate or to expand. For those from Colorado, the expectation is respondents who are against the status quo will be less likely to prefer state regulators than respondents who are for the status quo. For the issues in this paper with broader externalities, there is no difference in who currently regulates between Colorado and Texas. Therefore, the expectation is individuals who are against the status quo will be less likely to prefer state regulators than individuals who are for the status quo.
For each pair of issues, this paper completed two analyses: First, the marginal effects of policy preference on the probability of regulator preference. Second, the marginal effects controlling for state.
58


Marginal effects of policy preference for Issues with localized externalities: Setbacks and Nuisance issues. Figure 3.4 shows the marginal effects of all respondents policy preference toward fracking on their preferred level of government to regulate the issues of setbacks and nuisances. Each series in the figure represents a preferred level of government to regulate (local, state, or federal). The Y-axis represents the respondents probability of selecting the level of government. The X-axis represents the respondent policy preference categories (stop/limit, continue, expand).
For the issue of setbacks, the results show respondents who desire fracking to be stopped or limited have a probability of 0.61 to prefer local regulators, 0.21 to prefer federal regulators, mid 0.18 to prefer state regulators. Conversely, individuals who desire fracking to continue at its current rate have a probability of 0.56 to prefer state regulators, 0.41 to prefer local regulators, and 0.04 to prefer federal regulators. Finally, results in Figure 3.4 show individuals who desire fracking to expand have a probability of 0.76 to prefer state regulators, 0.25 to prefer local regulators, below 0.00 to prefer federal regulators.
The trends are similar for nuisance issues: as respondents policy preference related to fracking moves from stop or limit, to continue, then to expand, their probability to prefer state regulators increases from 0.21 to 0.57. A notable difference is that individuals whose policy preference is for fracking to continue at the current rate prefer local regulation (Pr(local regulation) = 0.53) more than state regulation (0.42). In for both local issues, the probability for preferring federal regulators across all groups does not go above 0.21 and moves toward 0 as policy preference changes from stop/limit to expand.
59


SETBACKS
a
o
| 0.8
at
X
IX
§ 0.6 x
at
O
^ 0.4
x
D
O
x
& 0.2
at
x
0.76,
State Preference
NUISANCE
0.8
x
U
z
X
X
X
£ 0.6 x
O
^ 0.4
O
x
£ 0.2
-V
CL
Figure 3.4. Marginal effect of policy preference on regulator preference for localized
24
issues. 24
24 Margins over policy preference for different outcomes: margins, over(policyjpreference) pr(out(2)) level(90); margins, over(policyjpreference) pr(out(3)) level(90); margins, over(policyjpreference) pr(out(4)) level(90).
60


Marginal effects of policy preference between states for localized issues. Figure 3.5, shows the difference in respondents preference for local, state, mid federal regulators for setbacks and nuisance issues between those from Colorado and Texas. See Appendix A Table 5-12 for the differences between state marginal effects and their significance. For setbacks, the results show respondents from Texas are more likely to prefer local regulators than respondents for Colorado. However, respondents in both states who desire fracking to be stopped or limited are more likely to prefer local regulators than policy actors who desire fracking to continue or expand. Additionally, respondents from both states are more likely to prefer the state to regulate the issue if they desire fracking to continue or expand, than if the respondent desires fracking to be stopped/limited. While the results do show a significant difference between respondents from Texas and Colorado, the results do not support Hypothesis 2. For nuisance issues, there is no difference between respondents from Colorado and Texas in their preferred level of government for regulation. The results show similar trends with respect to a respondents preference for local and state regulator: respondent who desire fracking to be stopped/limited are more likely to prefer local regulators mid less likely to prefer state regulates than respondents who desire fracking to continue or expand.
While these results support Hypothesis 2 for respondents from Colorado, the general trends imply respondents who area against the status quo are generally against state-level regulation and those who are for fracking are generally for state-level regulation. Finally, given the ubiquitous low preference for federal regulators for local issues, and high preference for local regulators implies some directionality to the regulator preference based on the nature of the issue, rather than which level of government currently regulates the issues.
61


Predictive Margins: Setbacks
Local Regulator State Regulator Federal Regulator
Colorado Texas
Colorado Texas
Predictive Margins: Nuisance Issues
Local Regulator
------- Colorado
--- Texas
State Regulator
------- Colorado
--- Texas
Federal Regulator
------- Colorado
--- T exas
Figure 3.5. Marginal effect of policy preference, by state, on regulator preference for localized issues.
62


Marginal effects of policy preference for issues with broad externalities: Air and water quality. Figure 3.6 shows the marginal effects of policy preference on regulator preference. The preferences for who regulates are similar across both issues: respondents who desire fracking to be stopped or limited also prefer federal regulators over state or local regulators. Additionally, results show respondents who desire fracking to continue or expand also prefer state regulators over federal or local regulators. For example, for the issue of air emissions, the respondents probability for preferring state regulators moves from 0.29 if they desire fracking to stop or be limited, to 0.70 if they desire fracking to continue, and to 0.86 if they desire fracking to expand. Given that both issues are regulated at the state in Colorado and Texas, these results align with Hypothesis 2. Further, the relatively higher preference for federal regulators to local regulators for these shows a similar directionality in the preferred alternative regulator for issues with broad externalities as seen with the localized issues.
63


PR (REGULAl'OR PREFERENCE) PR (REGULATOR PREFERENCE)
AIR QUALITY
0.8
0.6
0.4
0.2
0.86,
State Preference
0.70

0.53
Ok /'
0.29 -|
u' .022
0.18
0.07
STOP/LIMIT CONTINUE
Oil,
Federal Preference
0.04,
Local Preference
EXPAND
WATER QUALITY
0.8
0.6
0.4
0.2
Figure 3.6. Marginal effect on regulator preference for air and water quality over policy preference.
64


Marginal effects of policy preference between states. Figure 3.7 shows the marginal effect of the state on the respondents regulator preference for broader issues of air mid water quality. For both air mid water quality issues, results show significant differences between respondents from Colorado and Texas in the probability of preferring state and federal regulators and no difference between Colorado and Texas respondents on their preference for local-level regulators. Given the regulations for air mid water quality are both held at the state in Colorado and Texas, there is no theoretical expectation or explanation for this difference.
Respondents from Colorado are more likely to prefer state regulators than local or federal regulators across all policy preference categories for both air and water quality issues. Even though the Coloradoan respondents who desire fracking to be stopped/limited are less likely to prefer state regulators than those who desire fracking to be continued or expanded, these results to do not align with Hypothesis 2. Indeed, the shape of the curve between across the local, state, and federal marginal plots for the stop/limit group should be either U-shaped or a diagonal line, showing either the probability to prefer local or federal regulator to be greater than the probability to prefer state regulators. Texan respondents, however, show the hypothesized preference profile based on their policy core belief of whether fracking should be stopped/limited, continued, or expanded and the status quo of state-level regulators. Texan respondents who wish fracking to be stopped/limited are most likely to prefer federal regulators and least likely to prefer local regulators for both air and water quality issues. Texan respondents who wish fracking to continue or expand are more likely to prefer state-level regulators over local and federal regulators for the issue of water quality, and more likely to prefer state-level regulators than local level regulators, and equally likely to prefer federal regulators for air quality issues.
65


Overall, the results of the marginal effects analysis for the two broader mid localized
issues by state give mixed support for Hypothesis 2 that those who are against the status quo would be less likely to choose the level of government who currently regulates the issue. The nuances observed in the marginal effects plots on preferred regulator between localized and broader issues, and those between states, indicate the nature of the issue and the relationship between pro and anti-status quo to regulators is a potential driver for preference.
66


Predictive Margins: Air Quality
Local Regulator
------- Colorado
Texas
State Regulator
Federal Regulator
00 _
o -
h------------1-----------r~
stop/limit continue expand
------- Colorado
---- Texas
Predictive Margins: Water Quality
Local Regulator
------- Colorado
--- Texas
State Regulator
Federal Regulator
------- Colorado
--- Texas
Figure 3.7. Marginal effect on regulator preference for air and water quality issues; state and policy preference interaction.
67


Interaction of Deep Core mid Policy Core Beliefs on Secondary Beliefs
The nuances identified in the results indicate this papers hypotheses did not anticipate the policy status quo correctly. Results indicate a policy actors policy core belief -their preference toward fracking affects their secondary belief their preference for which level of government should regulate, but it is not explained by which level of government currently regulates. Returning to the ACFs hierarchical belief system, the final analysis examines the interaction between a deep core belief (government attitude) and a policy core belief (preference related to fracking policy) on a secondary belief (regulator preference) for each issue.
Figure 3.8 and Figure 3.9 show the marginal effects of government attitude and policy preference on regulator preference for local issues. Figure 3.10, and Figure 3.11 show the marginal effects of government attitude mid policy preference on regulator preference for broad issues. For localized issues, the effect of deep core beliefs (indicated by the slope of each line) on the probability of preferring a specific regulator is insignificant The driving factor for regulator preference is the policy core belief of policy preference. For broad issues, deep core beliefs play a strong mediating role. For example, examining the issue of air emissions, the probability of an individual who desired fracking to be stopped to prefer local regulators is 0.4 when their government attitude score is -1.5 (desiring less government involvement) and nearly 0 when their government attitude score is 2 (desiring more government involvement). Similarly, the probability of an individual who desired fracking to be stopped to prefer federal regulators is 0.22 when their government attitude score is -1.5 and over 0.70 when their government attitude score is 2.
68


Predictive Margins over Gov. Attitude: Setbacks
Local Regulator
State Regulator
Federal Regulator
u>

CO
(O

-1.5-1 -.5 0 .5 1 1.5 2 Gov. Attitude
------- stop/lim it
* continue expand
-1.5-1 -.5 0 .5 1 1.5 2 Gov. Attitude
--------- stop/lim it
* continue .... expand
4- 4 -4 A- 4 -4-4- -A
h---1iii-----1ir
-1.5-1 -.5 0 .5 1 1.5 2 Gov. Attitude
-------- stop/lim it
* continue expand
Figure 3.8. Marginal effect of government attitude and policy preference on regulator preference for setbacks and nuisance issues.
Predictive Margins over Gov. Attitude: Nuisance Issues
Local Regulator State Regulator Federal Regulator
ft:
**
tn
A- .
-1.5 -1 -.5 0 .5 1 1.5 2 Gov. Attitude
------- stop/lim it
* continue expand
Gov. Attitude
-------- stop/lim it
* continue expand
-1.5-1 -.5 0 .5 1 1.5 2 Gov. Attitude
-------- stop/lim it
* continue expand
Figure 3.9. Marginal effect of government attitude and policy preference on regulator preference for setbacks and nuisance issues.
69


Predictive Margins over Gov. Attitude: Air Quality
Local Regulator
Gov. Attitude
-------- stop/Iim it
* continue expand
State Regulator
Gov. Attitude
-------- stop/Iim it
* continue expand
Federal Regulator
Gov. Attitude
-------- stop/Iim it
continue expand
Figure 3.10. Marginal effect of government attitude and policy preference on regulator preference for air emissions and water quality.
Predictive Margins over Gov. Attitude: Water Quality
Local Regulator
Gov. Attitude
-------- stop/Iim it
continue expand
State Regulator
Gov. Attitude
-------- stop/Iim it
continue * expand
Federal Regulator
Gov. Attitude
-------- stop/Iim it
* continue expand
Figure 3.11. Marginal effect of government attitude and policy preference on regulator preference for air emissions and water quality.
70


Conclusion
The analyses in this paper show no single variable explains an individuals preference for which level of government should regulate fracking-development related issues. Overall, the analysis did not completely confirm either Hypothesis 1 or Hypothesis 2 (Table 3.2). While the results of the broader issue models gave support to the papers two hypotheses, the models predicting regulatory preference for localized issues did not. However, the alignment of the results of the models predicting regulatory preference for broader issues with respect to Hypothesis 2 how the status quo affects an individuals regulatory preference is likely due to the respondents general view of state, local, and federal regulators, rather than which regulatory body currently regulates specific issues. Indeed, in Colorado and Texas at the time of this research the state regulatory bodies were supportive of oil mid gas operations. This general view of state-level regulators is seen in the results of each model and is consistent regardless of the post estimation effects. Figures 3.5 through 3.11 shows individuals are more likely to prefer state level regulators across all issues if they have the policy preference that fracking continue or expand. The figures also show individuals are less likely to prefer state level regulators if they have the policy preference for fracking to be stopped or limited.
The results also show the policy core beliefs more consistently constrain secondary beliefs than deep core beliefs. An individuals deep core beliefs have no effect on regulator preference for localized issues, but they have a significant influence on regulator preference when issues have broader externalities. The respondents policy core belief on whether fracking should be stopped/limited, continued, or expanded had the most significant effect on the respondents secondary belief of their preferred level of government to regulate an issue. Of the four issues examined in this research, the respondents stance on fracking significantly
71


impacted their preferred level of governmental regulator. However, results show the relationship of the respondents policy core belief and the status quo as measured by which level of government currently regulates the issue is not supported. While results indicate respondents who desire fracking to be stopped or limited also prefer local-level regulators, the results also indicate they maintain this preference regardless of whether the issue is currently regulated by the state or local governments. Additionally, while the regulatory structure had the expected effect on preference for who regulates a localized issue of setbacks more preference for local regulators it did not on nuisance issues. This may be that the Rule 800 in Colorado allows local governments to exempt themselves from the state rules related to nuisance issues but the rule does not allow them to create their own rules, like in Texas.
72


Table 3.2. Results Summary.
Hypotheses Localized Issues Broader Issues
Hypothesis 1: Policy actors who believe governments should be involved less in daily life prefer lower levels of government to regulate oil and gas issues. Not Supported Supported
Hypothesis 2: Policy actors whose beliefs do not align with the status quo are more likely to prefer regulators at levels of government that are different than where they are currently administered. Limited support Supported
73


Discussion and Limitations
This research shows an individuals ideology, policy preference, mid political context impact an individuals preferred regulator for different issues related to hydraulic fracturing-based oil mid gas development. Additionally, this research shows the affect that these factors have on mi individuals preference for who should regulate varies depending on the nature of the issue in question. While these findings fit well with the ACF (i.e., several beliefs mid strategies shape the preferences of a policy actor) additional theory is needed to explain the variation observed in the models on an individuals preference for regulator.
The interaction effect of deep core and policy core beliefs on secondary beliefs observed in this research is worth exploring further. Why, for example, is the effect of a deep core belief (i.e., an individuals attitude toward government involvement in daily live) on the secondary belief (i.e., preferred regulator) so strong for issues with broader externalities (Figure 3.9), but the effect of deep core beliefs evaporates for issues with localized externalities (Figure 3.10)? Indeed, a dynamic in the ACF that needs further theory building is how the myriad of an individuals beliefs translate down the hierarchy and eventually manifest into policy goals on which the individual acts (Weible, Sabatier, & Lubell, 2004). In an early evaluation of the ACF, Jenkins-Smith and Sabatier (1994) jettisoned the idea that abstract beliefs constrain specific ones (p. 196), but they maintain beliefs continue to be hierarchical in nature. Weible, Sabatier, and Lubell (2004) provide empirical evidence that higher level beliefs inform lower level beliefs, but they found the line of influence between core and secondary beliefs may neither be direct nor include policy core beliefs. The results of this research provide additional empirical evidence for the linkage between higher and lower-level beliefs, but also the limits of deep core beliefs on secondary. Focusing on the
74


nature of the issue, such as the level of abstraction of the issue, may offer some explanation. For example, air mid water issues are broad mid individuals may interpret those issues in a number of ways and so their deep core beliefs help them develop their preference for who regulates. But the setback distances and nuisance issues are very tangible, so deep core beliefs are not needed as a heuristic in choosing their preference, and so their policy core belief has more influence influential in their preference of who should regulate.
Another observation to consider further is why those who desire fracking to expand consistently prefer state regulators and those who prefer fracking to be stopped or limited prefer either local or federal regulators. One potential explanation for the relative preference for or against state regulators found in the political conflict and venue shopping literature is the presence of iron triangles and captured regulators. For example, across all four issues, respondents who desired fracking to be stopped or limited were the least likely to prefer state regulators and respondents who desired fracking to be expanded were most likely to prefer state regulators. This relative likelihood to prefer state regulators was consistent across the policy core belief related to fracking, regardless of the respondents deep core belief (i.e. their attitude toward government) or if they were in the Colorado or Texas subsystem. Given that state regulators were considered by interest groups to be sympathetic to oil mid gas development, this may impact the preference for who regulates across all issues. A similar response is identified in the venue shopping literature: interest groups will seek out venues who are sympathetic to their cause (Pralle, 2003; Holyoke et ah, 2012; Constantelos, 2010) and avoid venues where opponents are present (Hall & Daerdorff, 2006; Hojnacki &
Kimball, 1998; Ley & Weber, 2015). However, this does not explain another dynamic on display in the analysis: those who wish for an expansion of fracking have even more
75


preference for state regulators than those who desire for continuation of fracking at its current rate. This research considered both to be on the side of the status quo, but the results show that there is need to develop hypotheses related to the direction of change, rather than simply change/no change.
Leaving state preference aside, a final observation of note is that the respondents preference for who regulates appears to have some directionality (i.e., individuals prefer local regulators for the localized and federal regulators for broader issues), indicating other significant factors exist. A more nuanced hypothesis is needed to address the directionality of regulator preference. For example, hypotheses that take the nature of the good, or issue in this case, into account. Respondents may be examining the issue from a more practical or strategic level, such as the policy actors view on a governing bodys ability to address a problem. In this sense, the nature of the issue is driving which level of government is more efficient at its governance an individuals beliefs are affecting the cost on their preference (North, 1984; North, 1990).
As Buchanan mid Tullock (1962) described the optimal size of government, they saw issues with larger externalities better handled by larger or higher-levels of governments. Conversely, issues with smaller externalities are better handled by smaller governments. Buchanan and Tullocks (1962) argument is buttressed by Butler mid Macey (1996) and they named the idea the matching principle. Also in the 1960s, the idea of polycentricity arose through observation of how metropolitan areas have found solutions to multi-scale issues (V. Ostrom, Teibout, & Warren, 1961). V. Ostrom et al. (1961) argued a polycentric, or multilevel and overlapping, governance system is a necessity to manage events with a range of scales of positive or negative externalities. Otherwise, when a public agencys boundary of
76


control does not match the boundary of the event they lose their ability to regulate effectively (V. Ostrom et al., 1961, p. 835). Research on common pool resources provides empirical evidence of the importance of matching the governing boundaries to the resource mid its users (E. Ostrom, 1990). Larger common-pool resources are a challenge to manage with small scale, self-governing appropriators because of a mismatch in governance mid resource boundaries (E. Ostrom, 2005, 283). Therefore, a policy actor who is practically, or efficiently minded for solving problems, may prefer local governments regulating smaller-scale issues related to oil mid gas development, and state or federal levels of government for regulating larger-scale items. But then, their attitudes toward government or a specific regulator mediate that belief, driving some at extreme ends elsewhere. Given that this research only examined four specific issues, further research is needed in this area to confirm or deny the results.
These alternative interpretations highlight some of the limitations of this study. For example, the inability of the theories used to explain and incorporate how the policy actors views of local, state, and federal regulators in the papers hypotheses or building a stronger connection between the nature of the good mid the policy actors preferred regulator. And more practically, the papers focus on only four issues limits the generalizability of the findings. However, as with most limitations mid unanswered questions, they also guide the direction of future research. As such, future work should include a broader range of issues with clear attributes, such as size of externality, incorporate theories on how prevailing institutional arrangements affect choice, and finally theories on how broader strategies of dominant and minority coalitions may play into their preference for who should regulate an issue. Indeed, one would expect different results in subsystems defined by a substantive topic that is not as contentious as fracking.
77


CHAPTER III
POLICY ACTORS VENUE SHOPPING PATTERNS DURING NEW YORKS
FRACKING DEBATES
Introduction: Venue Shopping
No other strategic decision may be as critical for a policy actor during a contentious debate than where to debate the policy issue. Contentious policy debates involve multiple strategic decisions mid actions made by policy actors (e.g., interest groups, governmental representatives, scientists, or the media). Strategic decisions include choosing with whom to ally, how to leverage focusing events, how to develop issue frames and narratives, which policy solutions to propose, mid where to debate mi issue. The decision of where to hold a policy debate, defined as venue shopping, is a specific tactic used by policy actors to expand or contain political conflict (Baumgartner & Jones, 1993; Schattschneider, 1975; Sabatier & Jenkins-Smith, 1993; Pralle 2003). Venue shopping also partially determines which governmental decision makers are active in addressing a policy issue, and to whom policy actors will advocate for their policy preferences. Indeed, policy actors select a venue, in-part, based on the decision makers associated with the venue (Holyoke, Brown, & Henig, 2012; Weber & Ley, 2015). Venue selection is a component of opening a new venue to a current policy issue, which is a mechanism for policy change (Kubler, 2001; Norhstedt, 2011). Finally, venue shopping requires the use of a policy actors limited resources, and so contains inherent risk because in every policy debate, there is a winner, and there is a loser. 25
25 In this research, the term venue is limited to only governmental venues. In the broadest use of the word, venues could include media and other public outlets for policy debates.
78


One reason policy actors venue shop is to expand political conflict. When a policy actor engages new set of governmental decision makers, they draw fresh attention to an issue (Baumgartner & Jones, 1991; Schattschneider, 1975). To affect such an exp mis ion, policy actors may re-frame a policy issue, find specific aspects of the issue that can be addressed by a different venue, or choose a venue that is sympathetic to their point of view and problem definitions (Baumgartner & Jones, 1993). When policy actors find a sympathetic set of decision makers at a new venue, they can more easily break the established policy images or iron triangles (Baumgartner & Jones, 1993). Indeed, as part of a conflict exp mis ion strategy, venue shoppers intentionally increase the attention paid to a policy issue. Once decision makers and the public turn their attention to an issue, a positive feedback loop, further increasing attention to the issue. Scholars show, increased attention is a necessary condition for major policy change (Baumgartner & Jones 1993; Schattschneider, 1975).
Another reason policy actors venue shop is to contain conflict. Policy actors will select venues with decision makers at venues who are supportive of the status quo to contain a political conflict. They may achieve a similar effect by lobbying to keep debates within venues previously involved in the policy issue (Pralle, 2003). By maintaining the traditional venues involved in an issue, policy actors can minimize new attention to the issue. This maintains current policy images and can block the very policy change efforts described above (Pralle, 2003). Venue shopping is therefore both a possible strategy used by policy actors to either instigate or oppose policy change through what scholars describe as overcoming or engaging institutional friction (Baumgartner & Jones, 1993; Baumgartner & Jones, 2005; Weible, Heikkila, deLeon, & Sabatier, 2012; Moe, 2015). As Papillons (2011) study of multi-level venue shopping shows, venue shopping can affect the very institutional
79


arrangements that create friction or other path dependencies faced by policy actors engaged in contentious policy debates.
Despite the effect venue shopping can have on policy debate outcomes, not every policy actor engages in venue shopping (Buffardi et al., 2015). This is because it is a costly endeavor. Policy actors must acquire mid use resources such as attention, time, money, and political connections and capital to engage a venue to advocate their policy position (Sabatier & Weible, 2007).
Given the cost mid the potential impact venue shopping may have on the outcomes of contentious policy debates, policy scholars can learn about strategic policy actor behavior by examining a venue shoppers choices. Indeed, scholarship on venue shopping is experiencing a reemergence with calls for, and development of, new theoretical models and quantitative models for such reasons (e.g., Ley & Weber, 2015; Constantelos, 2010; Holyoke et al., 2012; Ley, 2016; Beyers & Kerremans, 2012). However, this research limits our understanding of policy actor strategic behavior in two ways. First, the above scholars do not include the range of policy actors involved in policy debates; rather they focus on a single interest group (e.g Holyoke et al., 2012; Ley, 2016; Buffardi et al., 2015; Beyers & Kerreman, 2012; Constantelos, 2010).26 Second, the vertical or multi-level venue shopping research designs are complex and include venues in multiple states or the nation as a whole, which draws theoretical focus away from the venue-policy actor-relationship and onto institutional features affecting policy actor behavior (e.g., Beyers & Kerreman, 2010; Constantelos,
2010). Further, while the work is intended to examine how large institutional features affects
26 Holyoke et al. (2012), for example, surveyed charter schools in three states boards. Leys (2016) case study focused on industry groups in Oregon. Buffardi et al. (2015) examined nonprofits in Seattle. Beyers and Kerremans (2012) studied NGOs, business organization, and labor associations in the European Union. Constantelos (2010) studied trade, business, or professional associations in Ontario, CA and Michigan, USA.
80


the openness of venues, other institutional arrangements may exist (e.g., norms) that could affect political strategies like venue shopping decisions. Our understanding of policy actor behavior mid venue selection can benefit from research that is designed isolate the inquiry more closely policy actors and how their perception of the policy venue.
This paper builds on the venue shopping literature by applying the Advocacy Coalition Framework (ACF) to examine the venue choices of policy actor. Not only does the ACF provide theoretical guidance on key policy actor attributes affecting their actions, but its theoretical mid analytical tools address the two limitations in venue shopping research. First, venue shopping research typically focusses only on interest groups, while the ACFs advocacy coalition includes all policy actors involved in venue shopping. Second, venue shopping research often examines patterns across state and national boundaries. This research uses the ACFs subsystem to set analytical boundaries around a single state. This boundary simplifies the institutional features that could affect the study of venue choices, but still includes venues at multiple levels and branches of government (Sabatier, 1988). The research uses survey data from policy actors involved in New Yorks statewide fracking policy debates, and develops separate ordered logistic regression models to test two classic venue shopping questions
1. What factors influence the total number of venues shopped by a policy actor and;
2. How does a policy actors perception of a policy-making venue affect their shopping frequency at that venue?
The remainder of the paper will first outline the theoretical underpinning of the research through a review of ACF. Then, the paper develops the hypotheses for each research
81


question using insights from venue shopping literature mid the ACF. Following the hypotheses section, the paper describes the research setting New Yorks oil and gas policy subsystem. Next, the paper operationalizes each variable mid explains the regression models used for each research question. Next, the paper describes the results of the two models. The paper concludes with a discussion of the results and a reflection on the venue shopping literature.
Theoretical Foundations
The Advocacy Coalition Framework (ACF)
Two constructs within the ACF are central to its theories of policy change mid policy actor behavior in contentious contexts: the policy subsystem and the advocacy coalition. A policy subsystem is defined by a geographic area, a policy topic, and the policy actors within the geographical area involved in that topic. The policy subsystem construct simplifies analyses of complex policy process because it allows an analyst to define internal versus external influences on the policy subsystem and who qualifies as a policy actor. By focusing on a policy subsystem, scholars can identify and perhaps include multiple governmental decision-making forums. This is an improvement on typical venue shopping research that looks at a targeted venue or limited set of venues. In this research, the subsystem boundary is drawn around a single state. Therefore, the analysis includes decision-making venues at local and state levels of government mid across each branch of government. This provides horizontal mid vertical variation in venues and reduces the larger institutional features (e.g.,
82


different state or national rules on lobbying) which could complicate the analysis on venue
27
shopping decisions.
The ACF identifies a second construct: the advocacy coalitions. These advocacy coalitions operate within subsystems. This construct of the ACF simplifies analyses at the subsystem level of analysis. The advocacy coalition construct recognizes the multitude of policy actors involved in policy change (e.g., interest groups, the scientific community, and the media, and individuals from all levels of government). However, it does not require the researcher to identify or examine each policy actor. The ACF uses the boundedly rational individual, and insights of group behavior from the policy network theory to inform the advocacy coalition concept (Sabatier, 1988; Sabatier & Weible, 2007). The ACF assumes that individuals use their beliefs as heuristics to simplify their understanding of new information mid decision-making. As such, an individuals beliefs drive problem definitions and policy goals, and help determine who is an ally or adversary within a policy debate (Sabatier, 1988). Further, to overcome individual physical and cognitive limits, policy actors form coalitions to share resources and coordinate political activities to influence policy decisions. An advocacy coalition is therefore a broad network of policy actors with similar policy goals, who choose to act collectively to increase their ability to influence decision makers.
With respect to venue selection, the ACF posits that advocacy coalitions strategically engage governing sovereigns to ensure their decisions align with the coalitions beliefs 27
27 The ACF does not provide guidance on differentiating between available venues. Sabatier (1988) discusses how multiple venues exist, but because debates move from one venue to another over time, using the subsystem as the unit of analysis captures this movement without getting in to the details. The importance of different venues and the activities around venue selection and engagement are acknowledged as important (Sabatier & Jenkins-Smith, 1993 chapter 10).
83


(Sabatier, 1988). Although the term governing sovereign is used to describe the targets of advocacy coalitions, the implication is that policy actors in the coalitions aim to influence a set of decision makers who operate within one or more venues. The ACF neither describes nor explains the advocacy coalitions selection process between one governing sovereign and another. However, contemporary ACF research has highlighted the importance of such a selection process (Nohrstedt, 2011). Therefore, this research applies theories that are compatible with the ACF to develop hypotheses to explain the policy actors choice to engage with governmental venues. Specifically, venue shopping-related ideas from the Punctuated Equilibrium Theory (Baumgartner & Jones, 1993) and other related venue shopping scholarship (Pralle, 2004; Holyoke, Brown, & Henig, 2012; Constantelos, 2010) are applied to build hypotheses on the venue choices of policy actors. In all, the ACF assists this research on the strategic activity of venue shopping by setting up an analytical frame that includes all policy actors mid defines the subsystem boundaries, but traditional venue shopping theories are needed to develop specific hypotheses related to its research questions.
Hypotheses Development
Research Question 1: What factors influence the total number of venues shopped by a policy actor?
One way to achieve policy change is through political conflict expansion, which occurs when policy actors draw more people (i.e., decision makers, the public, interest groups, etc.) into the debate (Schattschneider, 1975). One way policy actors achieve conflict expansion is through venue shopping. When policy actors venue shop for this purpose, they strategically select decision makers who they believe will agree with their side of a political 28
28 Nohrstedt (2011) found major policy change occurred after a new policy venue opened to the debate as a result of actions of the minority coalition members.
84


conflict (Baumgartner & Jones, 1993). Once a set of decision makers attention is on an issue, a positive-feedback cycle of information and attention can be initiated. This facilitates the spread of the policy issue to other policy making venues (Baumgartner & Jones, 1993).
A subset of venue shopping scholarship in US and Europe describes and explains differences in horizontal or vertical venue shopping activity. Horizontal activity includes movement between venues in different branches or domains of government within the same level of government (Holyoke, Brown, & Henig, 2012). Vertical or multi-level venue shopping activity includes movement between venues in different levels of government (Princen & Kerremans, 2008; Beyers & Kerremans, 2012; Holyoke, Brown, & Henig, 2012; Constantelos, 1996, 2007, 2010). In research on vertical and horizontal venue shopping, scholars have defined greater levels of shopping activity by the total count of venues shopped (Beyers & Kerremans, 2012) or number of levels contacted by an interest group (Constantelos, 2010). In short, scholars quantify the level shopping through counting the total venues shopped. As such, this research examines factors that influence the total venues shopped by a policy actor. Specifically, the research will include resources, policy actor type, and if the policy actors view of current policies related to policy topic of interest.
Resources. The ACF identifies a broad range of resources including finances, leadership, access to authority, access to scientific and technical information, and mobilizable supporters (Sabatier & Weible, 2007; Weible 2007). The weight or relative importance of each of these influences is yet to be determined (Jenkins-Smith, Nohrstedt, Weible, & Sabatier, 2014; Nohrstedt, 2011). However, the venue shopping literature theorizes that resources affect a policy actors ability to venue shop (McQuide, 2010; Pralle, 2003;
85


Constantelos, 2010; Holyoke et al., 2010). Therefore, the first hypothesis related to total venues shopped is:
Hypothesis 1: Policy actors with more resources will shop more venues than those with fewer resources.
Although both theory and select studies show a positive relationship between a high resource level and increased venue shopping activity, this hypothesis is of continued interest because some empirical studies of venue shopping activity downplay the effect of available resources. For example, the level of resources explanation provided little explanatory power when examining interest group activity in multi-level venue shopping settings (Constatnelos, 2010; Beyers & Kerremans, 2012).
Policy actor type. The tendency of a policy actor to engage in political activity may depend on their organization type. For example, scientists may produce information related to a policy issue and, therefore, be appropriately considered as policy actors within the subsystem, but scientists may not engage with decision makers in the same way or frequency as interest groups. Likewise, governmental representatives at one level of government, or branch, may testify or advocate their preferences to another governmental decision-making body, but have constraints on when or how they can act. Indeed, interest groups, by definition, advocate for policies to align with their positions (Baumgartner & Leech, 1998). Therefore, the expectation of this paper is that interest groups will have a greater amount of shopping activity than other policy actors identified in the policy subsystem.
Hypothesis 2: Policy actors associated with interest groups will shop venues than policy actors associated with non-interest groups.
86


Beliefs of the policy actors. Policy actors are belief-motivated individuals, in that they seek to see their beliefs translated into policies through political activities, such as engagement with governmental decision makers (Sabatier, 1988). Therefore, if a policy actor determines the status quo policies do not align with their views, they are motivated to expend resources and advocate for change. Conversely, if a policy actors beliefs align with the status quo policies, they may be less motivated to seek change. As venue shopping theory posits, an individual seeking change will attempt to expand the political conflict to facilitate policy change in their favor (Baumgartner & Jones, 1993; Pralle, 2006; Pralle, 2003). Therefore:
Hypothesis 3: Policy actors whose beliefs deviate from the status quo will shop more venues than policy actors whose beliefs align with the status quo.
Control Variable: Organizational Focus. An organizations mission impacts its venue selection (Constantelos, 2004; 2010). While Constantelos (2004; 2010) found interest groups shopping outside of their stated focus or mandated jurisdictions, the logic remains sound that an organization created to influence a particular level of policymaking would shop at that level more frequently than others would. For example, if an organizations mission is to advocate a policy issue within a specific state within the United States, their organizational focus would be the state-level. Alternatively, if the organizations mission describes local activity and community building as their preferred mode to address a policy issue, then their organizational focus would be local. Depending on the context policy debate, organizations focused at one level of government may be more involved than others may. While no expectation is set in this research, the organizational focus is included as a control variable.
87


Full Text

PAGE 1

POLICY ACTOR BELIEFS AND BEHAVIOR S IN CONTENTIOUS POLICY DEBATES: EXAMINING POLICY ACTORS WITHIN THE STATE WIDE FRACKING SUBSYSTEMS OF COLORADO, TEXAS, AND NEW YORK. by SAMUEL BALLOU GALLAHER B.S., Oregon State University, 2004 M.P.A, University of Colorado Denver, 2011 A thesis submitted to the Faculty of the Graduate School of the University of Colorado in partial fulfillment Of the requirements for the degree of Doctor of Philos ophy Public Affairs Program 2017

PAGE 2

ii 2017 SAMUEL BALLOU GALLAHER ALL RIGHTS RESERVED

PAGE 3

iii This thesis for the Doctor of Philosophy degree by Samuel Ballou Gallaher has been approved for the Public Affairs Program by Tanya Heikkila Chair Christopher Weible Chair Benoy Jacob Rick Feiock Date: July 29, 2017

PAGE 4

iv Gallaher, Samuel Ballou (Ph.D., Public Affairs) Policy Actor Beliefs and Behavior in Contentious Policy Debates: Examining Policy Actors Within the Statewide, Fracking Subsystems of Colorado, Texas, and New York. Thesis directed by Professor Tanya Heikkila and Professor Christopher Weible ABSTRACT The goal of this dissertation is to address three areas in the policy process literature translate into secondary beliefs. To do so, the research models the ef view of government in daily life and their policy belief towards fracking on their secondary belief of which level of government should regulate an issue. Second, the research examines a behavior called venue shopping. The activity level, and how their beliefs toward decision makers influence venue selection. Third, the research examines local g overnmental representatives as policy actors in a state level policy subsystem. Policy process research identifies local government representatives within advocacy coalitions, but little is known about how local governmental actors compare to other advocat es in the coalitions. The research uses the Advocacy Coalition Framework (ACF) as an analytical and theoretical foundation and applies other policy process and organizational theories to inform its hypotheses. I use multiple quantitative data modeling tech niques to explore each question. Data for the research is from original surveys of policy actors in state level hydraulic fracturing subsystems in Colorado, Texas, and New York. deep core and policy core beliefs significant ly influence their

PAGE 5

v secondary beliefs. However, deep core beliefs have a greater effect on secondary beliefs related to more abstract issues such as air quality, and less on more concrete issues, such as the distance a well should be from other structures The venue shopping models indicate policy actors who oppose the policy status quo shop more venues than those who align with the status quo. Additionally, the strongest indicator of which venue a policy actor shops is not their beliefs toward the decision makers, but their other shopping choices. Finally, analyses show local governments are a unique group within and across coalitions because of their network relationships and they align with one another on a set of policy core beliefs, but are also divided among pro and anti fracking coalitions on other policy core beliefs Overall this dissertation shows the ACF provides a theoretical and analytical frame to examine policy actor beliefs and behavior, but additional theories and sub groupings of policy act ors are needed to explain nuances in policy actor dynamics. The form and content of this abstract are approved. I recommend its publication Approved: Tanya Heikkila and Christopher Weible

PAGE 6

vi For Caleb He never gave up, so neither did I. For Laura Thank you for your unending love and support and patience. I love you. For Mom and Dad.

PAGE 7

vii ACKNOWLEDGEMENTS The success of this research and dissertation and the completion of the program in general is due to a vast amount of support. First and forem ost, my co advisors and mentors, Tanya Heikkila and Chris Weible. Thank you for your all you did to find resources and bring me into the world of academic research through conferences, presentations, roundtables, and workshops. Your approach to educ ation and transformed me from an inquisitive individual to a thoughtful social scientist. I also wish to acknowledge your contributions to this dissertation You both guided the research from beginning to end, including used and countless edits on each chapter While I may not have always received edits and suggestions with a smile, once the wrangling of thoughts and words was done, the prod uct was in better shape than before. Chris. Thank you for pushing me to become a better writer. Thank you to Rick Feiock an d Benoy Jacob for your efforts on my dissertation committee. Your thoughtful comments and questions pushed me outside of my theoretical and interpretive comfort zone. Thank you to my fellow cohort and WOPPR ites and visiting scholar friend Jarkko While we struggled in our own ways, and often in the isolation of our own research and work, we formed friendships and, on those wonderful occasions, got to relax with a good drink. Thank huge thank you to my fellow WOPPR researchers on the Sloan Grant. Jon Pier ce, Jennifer Kagan, Ben Blair, our weekly meetings with Chris and Tanya were a

PAGE 8

viii our help in collecting state level regulatory information years ago. That information became a deep well to draw upon when writing my research setting sections. A special thanks to David Carter. I am thankful for our m any fine c onversations and that I lear n ed how to annoy you. Your unrelenting work ethic and dedication to the craft was an inspiration. Thank you to the School of Public Affairs and the staff and faculty. The first day I entered the school, a few weeks prior to starting the MPA program, I met with Richard Stillman and shortly thereafter, Peter de Leon. Richard, your enthusiasm for unabashed learning and intellectual exploration was truly inspiring. Also, once you learned I hoped to find work as a research assistant you promptly introduced me to Peter. A meeting which I thought was a just a friendly introduction. To my surprise, by the time I had made my way back to the elevators, Peter handed me a job description to work for him and Chris on the Policy Studies Journal. And like that I was a grad uate assistant. Who knows what would have happened if not for your way of supporting students? Thank you also to Malcom Goggin, Brian Gerber, Todd Ely, and Benoy Jacob. I am grateful for your willingness work with me and let me experience new and exciting research and methods. Thank you to Rob Drouillard for your friendliness and ensuring we had much needed access to software and files, while number of deadlines you kept m e from missing and forms you reminded me to fill out. Thanks for your open door and cheerful conversation! Thank you to my friends and family who saw me less and supported me more over the last six years. While I missed a lot, the game nights, camping tri ps, hut trips, dinners, gatherings, bachelor parties, weddings, and holidays that I made kept me grounded and sane.

PAGE 9

ix Preston, Josh, Mikey, Heather, Ian, Odele, Andy, Jill, Morgan, Luke, Ben, Sarah, Gavin, Curtis, Kara, Heather, Robert, Clif, Alex, Mike, Eva Brian, Ingrid, Matt, Amanda, Joel, Sarah, Danielle, Nick, Karen, Nate, Kerry, David, Sarah, David, Val, Jeremy, Sarah, Amy, Katie, Alex, Tomoko, Marcus, Patricia, Laura. You are my village. Thank you, Mom and Dad, for never doubting my success, helping w ith food and plane tickets, and the gazillion questions every time I called. Thank you, David and Sarah, for your friendship and opening your home to me, and soon thereafter, to Laura. It is not always the case you get to call roommates friends. Of course graduate student human being. If your kindness, laughter, patience

PAGE 10

x TABLE OF CONTENTS CHAPTER I. INTRODUCTION ................................ ................................ ................................ ........ 1 Background ................................ ................................ ................................ ................... 1 Research Setting: Statewide Hydraulic Fracturing (Fracking) Debates ....................... 4 Literature Review: The Advocacy Coalition Framework ................................ ............. 8 Chapter Introduction ................................ ................................ ................................ ... 17 II. BELIEFS AFFECT THEIR SECONDARY BELIEFS ................................ .............. 23 Introduction: Who Regulates ................................ ................................ ...................... 23 Contributions and Map of the Paper ................................ ................................ ........... 26 Theoretical Arguments and Hypotheses Development ................................ ............... 27 Research Setting ................................ ................................ ................................ .......... 33 Methods ................................ ................................ ................................ ....................... 42 Analysis and Results ................................ ................................ ................................ ... 50 Conclusion ................................ ................................ ................................ .................. 71 Discussion and Lim itations ................................ ................................ ......................... 74 III. FRACKING DEBATES ................................ ................................ ............................. 78 Introduction: Venue Shopping ................................ ................................ .................... 78 Theoretical Foundations ................................ ................................ .............................. 82

PAGE 11

xi Hypotheses Development ................................ ................................ ........................... 84 Research Setting ................................ ................................ ................................ .......... 94 Methods ................................ ................................ ................................ ....................... 96 Analysis and Results ................................ ................................ ................................ 102 Conclusions and Limitations ................................ ................................ ..................... 117 IV. COMPARING THE BELIEFS OF LOCAL GOVERNMENTAL POLICY ACTORS TO THEIR INTEREST GROUP ALLIES ................................ ............................... 121 Introduction: Local Governments as Policy Actors ................................ .................. 121 Theoretical arguments and Hypothesis Development ................................ .............. 123 Research Setting ................................ ................................ ................................ ........ 128 Methods ................................ ................................ ................................ ..................... 130 Analysis and Results ................................ ................................ ................................ 134 Conclusions and Limitations ................................ ................................ ..................... 148 V. COMPARING THE RESOURCES, NETWORKS, AND POLITICAL ACTIVITIES OF LOCAL GOVERNMENTAL POLICY ACTORS TO THEIR INTEREST GROUP ALLIES ................................ ................................ ................................ ...... 153 Introduction: Local Governments as Policy Actors ................................ .................. 153 Theoretical Arguments and Expectations ................................ ................................ 157 Research Setting ................................ ................................ ................................ ........ 166 Methods ................................ ................................ ................................ ..................... 167

PAGE 12

xii Analysis and Results ................................ ................................ ................................ 172 Conclusions and Limitations ................................ ................................ ..................... 186 VI. CONCLUSIONS ................................ ................................ ................................ ....... 192 Summary of work ................................ ................................ ................................ ..... 192 Contributions ................................ ................................ ................................ ............. 197 Future Research ................................ ................................ ................................ ........ 204 REFERENCES ................................ ................................ ................................ ..................... 206 APPENDIX ................................ ................................ ................................ ........................... 226 A Chapter 2 ................................ ................................ ................................ .................... 226 Descriptive Statistics ................................ ................................ ................................ 226 Hypo thesis 1 tabular data ................................ ................................ .......................... 228 Margins For Interaction Between Policy Preference And State ................................ ........ 229 B Chapter 3 ................................ ................................ ................................ ................... 233 Descriptive st atistics for RQ1. ................................ ................................ .................. 233 Descriptive statistics for RQ2 ................................ ................................ ................... 234 Local Level Action ................................ ................................ ................................ ... 236 Alternative Model for RQ1 ................................ ................................ ....................... 237 C Chapter 4 ................................ ................................ ................................ ................... 238 Respondent descriptive on policy position and org type ................................ .......... 238 H1: ANOVA ................................ ................................ ................................ ............. 238

PAGE 13

xiii H1: Full Fisher Ha yter Table ................................ ................................ .................... 239 H1: Correspondence Analysis ................................ ................................ ................... 240 H2: Difference in Means and Extremism ................................ ................................ 241 D Chapter 5 ................................ ................................ ................................ .................... 244 Factor Analyses of Resources ................................ ................................ ................... 244 H1: Comparison of Resources ................................ ................................ .................. 246 H2: Primary and Secondary Activities ................................ ................................ ..... 248 H2: OLS Progression of primary and secondary activities. ................................ ...... 249 Network Size ................................ ................................ ................................ ............. 251 Network patt ern MCA ................................ ................................ .............................. 252

PAGE 14

1 CHAPTER I I NT RODUCTION Background policy decisions are powerful because they activate their ability to coerc e actions of and redistribute resources to individuals and organizations. The Policy process is o ne way in which scholars describe and explain how governments and interest groups make and change polic ies Depending on a resea policy process can be linear or something For example, a single policy may follow a general path from inception through implementation. However when the researcher examines bundles of polici es surrounding a topic the policy process can be something with neither a beginning nor an end, only an evolution. In either case, t he policy process is a messy construct full of individuals acting over a landscape defined by rules and in an environment known for disruptions that come in the form of new and unexpected information. As policy scholars unpack the policy process es around a topic a few of the big questions asked are process did the outcome and the outcome Two of the key elements used to explain those big questions are the beliefs and the behaviors of the individuals involved in the policy process. Indeed, while a multitude of institutions (e. g. rules, laws, or norms) govern th e policy process, the policy process is also a social process in which individuals with strong belief and, cognitive and physical limitations are synthesizing information and making decision s Policy scholars who examine policy actor beliefs and behavior s often do so in the context of contentious debates. This is because c onten tious policy debates provide a research setting with rich variation in both belief and behavior to explore and test hypotheses. Eight

PAGE 15

2 defining characteristics of contentious debates highlight t hese points First, contentious policy debates involve multiple advocacy groups working collectively based on shared interests for or against the policy or policies in contention (Tilly & Tarrow, 2007; Sabatier, 1988). Second, advocacy groups, d efined as coalitions, consist of a multiplicity of policy actors from local, state, and federal levels of government, nonprofit and for profit organizations, media, and the scientific community (Sabatier, 1988). Third, the coalitions of policy actors engag e in a wide variety of activities in and out of the policy making venues (Tilly & Tarrow, 2007; Sabatier & Jenkins Smith 1993; Kingdon, 1984) to either increase or contain the political conflict (Schattschneider, 1978). Fourth policy actors target policy venues to influence governmental de cision makers, and fifth, the venues are located in multiple levels and branches of government (Sabatier, 1988; Sabatier & Jenkins Smith, 1993). Sixth, competing advocacy coalitions use scientific and technical informatio n to bolster their own arguments and/or to debunk their opponents (Jenkins Smith, 1988; Sabatier & Jenkins Smith, 1993). Seventh, contentious policy debates related to a single topic often last for a decade or more and so the individual policy actors and p olicy venues involved in the debates are not constant (Sabatier, 1988). Finally, policy brokers may be present and persuade policy actors (both governmental and nongovernmental) to find a policy solution to end the debate (Sabatier 1988 ). 1 1 The policy broker is one who does not have interest in the outcome of the decision, but does wish for the debate to end and pushes for a decision or negotiation to be made Another identified role is the policy entrepreneur (Baumgartner & Jones, 1993; Rocherfort & Cobb 1994; Mintrom & Vergani, 1997). The entrepreneur could be considered a policy advocate and part of an advocacy coalition, or a decision maker, or policy broker. The ACF acknowledges there are policy actors who are not part o f coalitions, but do contribute to the policy debates within a subsystem (i.e., by making decisions within a policy venue). The policy broker and entrepreneur have similar traits in that they may or may not be part of a coalition and have a vested interest in a policy change occurring. Mintrom and actively seeks broad change within a domain or subsystem while a broker or activist seeks specific policy change. Maloney et al., & Olsson (2011) b oth discuss the insider activist, which is a type of policy advocate.

PAGE 16

3 Research on t he policy process has taught us a great deal about policy actor beliefs and behaviors. For example, the I nstitutional A nalysis and D evelopment F ramework examines how rules and norms influences individual and group behavior s (Ostrom, 2005) The Multiple Str eams Framework (Kingdon, 1984), Punctuated Equilibrium Theory (PET) (Baumgartner & Jones, 1993), and the agenda setting work of Schattschneider (1975) and Rocherfort and Cobb (1994) each provide valuable insights to how the policy actors advocating for pol icy change (or stasis) employ different strategies to exploit political opportunities and influence decision makers and the broader political conflict. Policy actors may strategically select a venue or change venues to expand political conflict or policy actors attempt to keep discussions at the current venue to contain the conflict ( Baumgartner & Jones, 1993; Schattschneider, 1975; Sabatier & Jenkins Smith, 1993; Schattschneider, 1975 ; Pralle, 2003). Indeed, interest groups will strategically seek out ven ues where they think they have the best chances of achieving their policy goals (e.g., Baumgartner & Jones, 1993; Pralle, 2003 ; Constantelos, 2010). The Advocacy Coalition Framework examines how tivation for action, and act as a heuristic to filter information and identify allies (e.g., Sabatier, 1988). Each of these theories and frameworks ha ve strengths and limitations i n their ability to explain policy actor behavior s or policy process outcomes I t is not the goal of this dissertation to highlight or address each of the gaps H owever the four proceeding chapters examine specific gaps in our understanding of policy actor belief or behavior in contentious policy debates. Each chapter of this rese arch uses t he Advocacy Coalition Framework (ACF), a prominent theory for describing and explaining policy processes in contentious settings, as its theoretical foundation. The ACF lays out a general logic for policy actor behavior within

PAGE 17

4 the policy process The ACF also provides researchers with clear conceptualizations of the research setting and the policy actors involved. Finally, researchers can apply multiple theories within the ACF. This research applies theories compatible with the ACF, such as the t heory of venue shopping to develop specific hypotheses when explain or questions The remainder of C hapter 1 introduces the contentious policy debate in which each empirical chapter is set: the state level hydraulic fracturing based oil and gas development debates in the United States. Then it provides a literature review of the ACF and key concepts from the ACF applied in this dissertation Last, it introduces the central the me and question of each independent, empirical chapter. Research Setting : Statewide H ydraulic F racturing (Fracking) D ebates One example of a contentious policy debate is the issue of oil and gas development that uses of h ydraulic fracturing aka fracking Technological advances in fracking and horizontal drilling that began in the 1980s have enabled the economic extraction of oil and gas trapped in porous rock substrates (e.g., tight sands or shale). I n the United States, the oil and gas industry used the se technologies t o expand its operations in the mid 2000s into known shale and tight sand formation as well as in previously unidentified deposits. By the late 2000s, the industry was experiencing a modern day oil and gas boom. Fracking based o il and gas o perations expanded in areas accustomed to the industry and into areas unfamiliar with development. A reas unfamiliar with oil and gas development included population centers, residences, and schools Further, industry found more opportunities to increase op erations in environmentally sensitive areas such as lakes, streams, state and national forests, and wildlife preserves. Because of the risks associated with fracking, and where it is

PAGE 18

5 being applied, multiple environmental and health groups and communities m obilized to oppose oil and gas development that sued fracking across the United States. However, due to the economic benefits, mineral owners, developers, and oil and gas industry groups also mobilized to support the industry 2 By 2007 and 2008, states regulators, such as the Railroad Commission in Texas and the Colorado Oil and Gas Conservation Commission, had updat ed their oil and gas development policies to incorporate the fracking process (Hydraulic Fracturing Information, 2012) In New York, in 2008, the Governor Paterson placed a moratorium on fracking until the New York Department of Environmental Conservation could update its supplemental Brown, 2011; NYDEC, 2011 ). In addit ion, local governments in Colorado, Texas, and New York had also engaged in fracking policy debates. As a result, some local governments made policies that promoted development, and other local governments made policies in opposition to development (Gallah er, 2015; Fracktracker website). 3 Throughout these debates, policy actors fought to get policies changed in their favor (Heikkila et al 2014). Opponents of oil and gas development argue d to stop or limit fracking because the processes negatively impact s the environment and public health and safety (Food & Water Watch, 2015; Gallaher et al., 2014; Heikkila et al., 2014b; Pierce et al., 2013). Proponents of development downplay ed the environmental and health concerns, and argue d to continue fracking because 2 While several risks related to hydraulic fracturing based development were not part of the actual process of hydraulic fracturing (e.g., the truck traffic used to bring water to well sites, or methane emissions from well debate the issue. Fracking is therefore the blanket word used in this work as shorthand for hydraulic fractur ing based oil and gas development. 3 Federal level activity is also present in the United States. For example, debates related to hydraulic fracturing related regulation are found in both congressional and regulatory venues (e.g., the Bureau of Land Manage ment and the Environmental Protection Agency). Federal activity is not included in this research because the dataset used suggests that federal actors and issues have a played a minor role in state level debates (as of 2013) (Gallaher et al., 2014; Heikkil a et al., 2014b; Pierce et al., 2013).

PAGE 19

6 development provide d significant economic and national security benefits ( In the Matter of Changes to the Rules of the Oil & Gas Conservation Commission of the State of Colorado to Consider Hydraulic Fracturing Disclosure Rules, 2011; Hassett & Mathur, 20 13; Heikkila et al 2014b; COGA, 2014). The political conflict and policy debates surrounding hydraulic fracturing based oil and gas development are like other contentious policy debates in many ways. First, two opposing coalitions are attempting to sway policy outcomes and these coalitions are made of broad range of governmental and non governmental actors (Heikkila et al., 2014). R epresentatives from eac h level of government, alongside environmental and industry interest groups, royalty owners, agricultural representatives, and concerned citizen groups are mobilizing and participating in policy discussions. Second, the policy actors on both sides of the d ebates use scientific and technical information to support their arguments (e.g., process updating their Supplemental Environmental Impact Statement beginning in 2009 throug h 2014). Third, the policy debates have occur red over long periods For example, the states where the two technologies were first employed have addressed hydraulic fracturing based development concerns with pol icy change since the early 2000 s For instance Garfield County Colorado created their Energy Advisory Board in 2004 made of industry, environmental, public, municipal, and county representatives, then in 2005 Garfield County signed a Memorandum of Understanding with the Colorado Oil and Gas Conserva tion Commission, industry, and environmental representatives in 2005 to evaluate water quality and potential impacts from drilling Similarly, in 2003 the town of Flower Mound, Texas passed ordinances to regulate distance of wells to other buildings, noise safety, and

PAGE 20

7 environmental impacts, which have undergone multiple major policy change processes since 2007 Additionally, b oth Texas and Colorado underwent major rule making processes beginning in 2007 and New York began updating its supplemental general environmental impact statement (SGESI) for oil and gas oper a tions These states also are re kindling policy debates from over 30 years ago (Gallaher 2014; Minor, 2014). Finally, the policy debates around hydraulic fracturing based development resemble othe r contentious debates in that the policy actors involved are engaging in a wide variety of activities to sway both the example, policy advocates are using media, holding protests, re defining the issue to gain broader support, mobilizing support from their association members, and engaging multiple policy making venues (Gallaher et al 2014; Heikkila et al., 2014a; Heikkila et al., 2014b; Pierce et al., 2013 ; Gottlieb, 20 12; Meyer, 2012; Brush, 2013 ). Previous research on state level fracking policy debates shows loc al government representatives ( e. g. officials from county and municipal governments) as part of the advocacy coalitions involved in state level politics (Heik kila et al ., 2014a). The advocacy coalitions identified in the statewide policy subsystem are engaging in numerous activities at multiple policy venues at the state and local levels. Other research on hydraulic fracturing politics highlights local governme ntal decision makers are engaged in within their own jurisdiction on policy issues related to road maintenance, land use, setbacks, and environment (Riley, 2007; Groundwater Protection Council & ALL Consulting, 2009). Given that the issue of fracking involves statewide contentious policy debates and a variety of activity of interest groups and governmental representatives, it is an appropriate policy context to

PAGE 21

8 explore policy actor beliefs and behavior. This dissertation exa mines frack ing policy debates within Colorado, Texas, and New York. Literature Review: The Advocacy Coalition Framework The Advocacy Coalition Framework (ACF) is built to describe and explain the actions of policy actors in contentious political contexts The focus is on how policy actors form advocacy coalitions to engage in the policy process and interact with governing sovereigns (i.e. decision makers in policy making venues). For example, The ACF is theoretically geared to answer questions in three areas: questions related to policy change questions related to coalition behavior policy oriented learning This section describes parts of the ACF relevant to the main concepts used in this dissertation to examine beliefs and behavior of policy actors Those concepts in clude the policy subsystem, advocacy coalitions policy actors and policy actor beliefs Policy actor beliefs are also used in this dissertation to compare groups of policy actors within a subsystem. In addition to beliefs, t his section introduces three other concepts used in this dissertation distinguish local government al policy actors from policy actors associated with interest groups Those include resources for political advocacy political activities, and networks While the ACF has many s trengths, and provides researchers with concepts and relationships between the concepts to explain the policy process, the ACF also has limitations. For example, it lacks some explanatory power on how advocacy coalitions choose among the available governin g sovereigns within a subsystem to deploy their

PAGE 22

9 advocacy activities. Additionally, even though the ACF is built upon a model of the individual, it does not engage in explaining or comparing individual level attributes (beliefs withstanding) such as resourc es, political activities, or networks. W hen these limitations arise in each individual chapter, this dissertation borrows ideas from other compatible theories to develop its hypotheses and expectations Specific ACF l imitations are identified as each chapter is introduced below. Policy Subsystem s and Advocacy Coalitions Sabatier (1988) the founder of the ACF, recognized that the study of the policy process and policy change needed to expand beyond a focus on single policy events and decision makers at a single venue. To do so, he proposed that scholars change their unit of analysis to the policy subsystem He argued that viewing policy processes a policy subsystem enables a broader view on policy proc esses and change in three ways. First a substantive policy topic, rather than an event or single decision is the focal point of a subsystem Second, the policy subsystem includes all policy activity within a geographic region and can include multiple pol itical jurisdictions Third, the subsystem view acknowledges all policy actors engaged in policy processes related to a specific topic ( e g scientists, interest groups, news media, and decision makers at all levels of government) and all policy making venues within geographic boundary (Sabatier, 1988). While the analyst could vary the geographic scope of the subsystem to change the level of granularity of analysis of the policy processes within, typical ACF studies examine subsyst ems bound by national, regional, or state boundaries. Therefore, once the analyst defines the policy subsystem they can then define internal versus external influences on the policy subsystem

PAGE 23

10 actor and identify the availabl e political venues through which political conflicts may play out The ACF assumes subsystems are largely independent of each other due to the time and resources required for a policy actor to specialize on a topic to engage in a single subsystem (Heclo, 1978). However, outputs from one subsystem can impact other subsystems In these instances, t he outputs from one subsystem are treated as external events on another subsystem depends on the top ic salience and proximity of the two subsystems (Zafonte & Sabatier, stronger when the policy issue of the two subsystems is similar than when subsystems share the sam e geographic boundary, but focus on different policy issues (Nohrstedt & Weible, 2010). For example, outputs from the Colorado water policy subsystem may inform policy debates in the New Hampshire water policy subsystem, but outputs from the Colorado water policy subsystem would likely have little effect on the Colorado child welfare subsystem. Additionally, policy actors within different subsystems may interact in the search for allies or new policy making venues (Zafonte & Sabatier, 1998). In their resear ch Lubell, Henry, and McCoy (2010) argue actions like venue shopping increase the interconnections between dissimilar policy issues because a single policy making venue may hold authority over both topics. Therefore, the decisions made on one policy topic could affect decisions made within the same venue on a separate policy topic. The interactions between different policy subsystems are not a focus of this research. Advocacy coalitions operate within policy subsystems. The advocacy coalition

PAGE 24

11 coalition conceptually recognizes the multitude of policy actors involved in policy processes (e.g., individuals from all levels of government, interest groups, the sc ientific community, and the media), but does not require every policy actor to be identified or examined. In other words, when researchers collect individual level data they can treat the data as representative points and aggregate the information to descr ibe beliefs and political behaviors of coalitions. The ACF uses a model of the boundedly rational individual and insights of group behavior from policy network theory to inform the advocacy coalition concept (Sabatier, 1988; Sabatier & Weible, 2007). The gathering and decisions of the individual are filtered through their personal beliefs and that these beliefs drive their policy preferences. Furthermore, the ACF argues that to overcome individual physical and cognitive limits, policy actors form coalitions to share resources and coordinate political activities. Policy actors coalesce into coalitions, in part, by identifying with whom they share similar policy preferences (Sabatier, 1988 ; Zafonte & Sabatier 1998; Sabatier & Weible, 2007). An advocacy coalition is therefore a broad network of policy actors from local, state and federal governments, interest groups, the scientific community, and the media These policy actors have individual belief driven goal s, but choose to act collectively to increase their ability to influence decision makers, and translate their beliefs into policies. The nature of interactions between advocacy coalitions within a subsystem ranges from cooperative to conflicting (Weible, 2 008). In policy subsystems with a contentious substantive topic there are typically two or three conflicting coalitions (Weible, Sabatier, & McQueen, 2009), but can range between one and five (Weible, Sabatier, & McQueen, 2009). In contentious subsystems, the coalition that maintains political control over policy decisions

PAGE 25

12 over extended periods is considered a dominant coalition and acts to keep the status quo or supports policy change s that are congruent with their beliefs and support their goals When th ere is a dominant coalition, the opposition, mobilized in one or more minority coalitions, seeks policy change to affect policy in ways that are congruent with their beliefs ( Sabatier & Jenkins Smith 1993; No h r stedt, 2010 ). Minority coalitions often seek allies from outside the subsystem or take policy debates to venues tha t differ from where debates are traditionally held in the subsystem (Fritschler 1983 ; Baumgartner & Jones 1993 ; Browne 1990 ; Worsham 1997). Policy Actors Policy actors within a subs ystem are individuals, usually professionally affiliated with an organization, involved in the policy area and dedicating at least some time to influencing either directly or indirectly the politics of the subsystem. In contrast, an individual who submits an official comment on a policy debate, participates in a protest, or votes on a law related to a policy topic is not necessarily a policy actor. In the ACF, policy actors differ from other citizens by the time they devote to an issue and the extent they s pecialize in the issue. A policy actor may play multiple roles within a subsystem: they can be a policy advocate and member of an advocacy coalition, they may be a decision maker within a governing body, or they may be a policy broker. There is no reason o ne policy actor may not take on multiple roles within the same subsystem at different times or in different situations. For example, in one situation a governor may act as a policy broker for a regulatory body attempting to develop policies where competing coalitions are deadlocked on details within the policy. In a second situation, the governor may be the decision maker being lobbied by the competing coalitions.

PAGE 26

13 Policy Actor Comparison Beliefs, Resources, Activities, and Networks Scholars compare advoc acy coalitions using four key attributes from the ACF: and differences of coalitions in a subsystem, rather than individual policy actors, those attributes can be applied to examine policy actors and other groupings of policy actors. I argue that beliefs, resources, political activities, and networks can also be used to describe an individual s or an as they can be section briefly introduces the four attributes, leaving the deeper discussions of how the attributes may differ across groups to the relevant empirical chapter. T Policy actor belief s are a key attribute of individual policy actor s and a foundational element of the ACF The ACF views a policy actor s beliefs as their motivation for acting within the subsystem and the basis for their policy preferences. efs using a three tiered hierarchical belief system At the highest level of the hierarchy are deep core belie fs, then policy core, and finally secondary beliefs. In this hierarchy, beliefs range from the abstract to the specific (Sabatier & Jenkins Smith, 1993; Peffley & Hurwitz, 1985). multiple attributes including the level of abstraction, how empirically based the belief is, the geographic scope of the belief with respect to the subsystem, and the level of difficulty to change the belief (Sabatier, 1988; Sabatier & Weible, 20 07; Weible, Sabatier, & McQueen, 2009). For example, a values. Deep core beliefs are considered constant and are not related to specific policy topics.

PAGE 27

14 Policy core beliefs are thought t o be subsystem wide and define priorities such as whose welfare matters most in the subsystem, the role of government (i ncluding which level of government should regulate ), problem identification and its seriousness at the subsystem level, and preferred po licy solutions (Sabatier & Weible, 2007; Jenkins Smith, Nohrstedt, Weible, & Sabatier, 2014). Policy core beliefs are considered difficult to change but may shift over long periods of time of a decade or more (Sabatier, 1998). The lowest level beliefs are secondary beliefs. Secondary beliefs are not subsystem wide and typically associated (Sabatier & Weible, 2007 p. 196). Secondary beliefs are the most malleable o f the three belief types, yet still resistant to change, and are more easily measured (Weible & Sabatier, 2006). Overall, individual beliefs guide problem perceptions and policy preferences and are the inception of an policy goals (Sabatier, moderate information processing and act as a cognitive heuristic to identify potential allies ( Scholz & Pinney, 1995; Sabatier, 1988) Beliefs as a C omparative A ttribute Not only are beliefs a defining characteristic of policy actors, but beliefs can be used to describe differences between and within advocacy coalitions. While the ACF theory and empirical evidence from applications of the ACF shows policy actors form coalitions with others who share similar policy core beliefs, there is also evidence that policy actors of the same advocacy coalition have var ying policy core beliefs ( Sabatier, 1988; Nohrstedt, 2010) ACF scholars attribute coalition level variation in policy core beliefs, suc h as policy preferences to differences in individual

PAGE 28

15 affiliation (Jenkins Smith & Claire, 1993; Nohrstedt, 2005; Nohrstedt, 2010). There are endoge neity issues with the organization affiliation argument (Sabatier, 1988), however there is evidence that an individual may have their beliefs or self interests coopted by their hese organizational level goals then influence a policy final policy preferences. For example, the policy preferences of governmental actors are influenced by their interest for continued public support, and that this interest supersedes their policy core beliefs (Nohrstedt, 2005; 2010). Resources A second attribute for comparing policy actors is their resources for political activity The ACF states that resources give coalitions capacity to plan and act on different strategies and support their information processing and learning (Sabatier & Weible, 2007; Howlett, 2009; Elgin & Weible, 2013). For example, resources provide policy actors or coalitions with the capacity to engage decision making venues (Holyoke, Brown & Henig, 2012 ). Further, w hen two or more coalitions are engaged in a political debate, they use resources to influence policy outcomes (Jenkins Smith, 1988). Resource categories include finances, leadership, access to authority, access to scientific and technical information, and mobilizable supporters (Sabatier & Weible, 2007; Weibl e 2007). Because a resources are an aggregation of individual organizational resources the same logic that resources are a key attribute of coalitions and that those resources allow coalitions to act can be applied to individual policy actors. Political Activities The ACF connects the resources and beliefs of coalitions to its strategies to influence decisions by governmental authorities (Sabatier & Weible, 2007, Fig 7.1, pg. 191). This dissertation examines specific political activities rat her than broad strategies. If strategies are a plan or method for achieving a goal over a short or long period

PAGE 29

16 of time, then political activities can be thought of as the discrete actions of political advocates used to implement a political strategy. For e xample, a coalition may venue shop as a strategy to achieve a goal of policy change (Pralle, 2003), but within a strategy, many activities may ensue such as lobbying, testifying, or making official comments during policy making processes, or mobilizing tro ops to engage the decision makers of the venue. While describing the activities of a policy actor or coalition may not reveal their strategy, it does highlight the discrete ways in which they are engaging in the policy process within the subsystem. Networ ks One way to describe t he nature of activities and relationships between policy actors is through the policy network literature. For example, a n advocacy coalition represents a network of individual policy actors who coalesce through shared beliefs and a desire for policy change and coordinate activities aimed at achieving a policy goal. Advocacy coalitions, however, are but one type of network that may be found within a subsystem. Advocacy coalitions are like ally and coordination networks described in t he policy network literature (Salisbury, Heinze, Laumann, & Nelson, 1987; Zafonte & Sabatier, 1998; Weible & Sabatier, 2005). Other networks found in the policy network literature include power, information and advice, and resource sharing networks (Weible & Sabatier, 2005). Each of these networks share a common theme related to resource control or exchange and has the potential to include policy actors who are identified in separate coalitions. Indeed, applications of the ACF that examine the interactions of disparate groups acknowledge that policy actors who are members of conflicting advocacy coalitions may interact outside of the policy debates. These policy actors interact to acquire resources or because of institutional or functional links that dictate their interactions (Zafonte & Sabatier, 1998). W hile the ACF focusses on coordination of individuals as they relate to forming advocacy coalitions, other

PAGE 30

17 coordination patterns exist among policy actors within the subsystem. This dissertation does not defi ne the other coordination patterns, but describes the networks of individuals in two ways: first it describes a they collaborate with, and second, the number of other actors with whom the individual collaborate s C hapter I ntroduction The research for this dissertation is broken into four empirical chapters. Each chapter is written as a stand alone publishable article. Therefore, some information in the theory and background on the issue of fracking is repeated across the chapters. For example, t he research in e ach empirical chapter of this dissertation applies the ACF to examine policy actor engagement in the policy process within state level fracking debates. Chapter 2 hierarchical belief system in two ways. First, it models the relationship between deep core, policy core, and secondary beliefs beliefs interact with one another (Jenkins Smith et al., 2 016). Using the fracking debates in Texas and Colorado as the backdrop, the chapter general attitude toward the role of government in daily life (deep core) and their normative policy preference related to fracking (policy core) affects their preference for which level of government should regulate issues related to fracking (secondary). Policy process scholars agree that the higher level, normative, deep core beliefs inform policy core and secondary beliefs (Jenkins Smith & Sabatier, 1994; Sabatier, 1998; Sabatier & Weible, 2007; Jenkins Smith, Nohrstedt, relationship of humans and the natural environment will inform their policy preferen ce on

PAGE 31

18 climate change, and potentially their secondary belief on addressing water use in their community. But, the degree to which deep core beliefs constrain policy core and secondary beliefs remains unclear (Jenkins Smith & Sabatier, 1994; Weible, Sabati er, & Lubell, 2004). 4 Following the same example, ACF theory and empirical evidence from models of the belief will constrain their preference for a policy tool t o manage local water use (i.e. a secondary belief). Chapter 2 seeks to address this limitation within the ACF regarding how an Second, Chapter 2 examines how the context of the policy debate imp act s policy preferences regulate an issue. The ACF acknowledges that the context s urrounding the policy issue, such as the nature of the good, the current rules in place, and physical attributes related to the Smith & Sabatier, 1994). For example, Jenkins Smith & Sabatier (1994) describe how air policy is affected by the fact that air quality is a collective good and the physical properties of the earth that impact air flow (pg. 180). Indeed, the logic of the ACF highlights that advocacy goals are strategic and a result of contextual issues surrounding an issue and the beliefs of policy actors within the advocacy coalitions (Sabatier & Weible, 2007; Jenkins Smith, Nohrstedt, Weible, & Sabatier, 2014). The theory of venue shopping is appli ed alongside the ACF to develop hypotheses related to how normative beliefs and the context of a policy debate affect specific policy preferences. 4 Researchers find it is operationally difficult to differentiate between policy core and secondary beliefs (e.g., Olson, Olson, & Grawronski, 1999).

PAGE 32

19 Chapter 3 Chapter 3 examines a strategic behavior of policy actors venue shopping in the fracking debates in New York. Venue shopping is the act of a policy actor engaging a policy making venue (e.g., a state legislature, or court, or a city council) to achieve their policy goals. The current venue shopping literature limits our understanding of policy actor strategic behaviors in two ways. First, the current studies do not include the range of policy actors in a subsystem; rather they typically focus on interest group s (e.g., Holyoke et al., 2012; Ley, 2016; Buffardi et al., 2015; Beyers & Kerreman, 2012; Constantelos, 2010). 5 Second, the vertical or multi level venue shopping research designs are complex and include venues in multiple states or across national boundaries, which draw theoretical focus away from policy actor behavior and onto institutional effects (e.g., Beyers & Kerreman, 2010; Constantelos, 2010). To address these limitations in the venue shopping literature, Chapter 3 applies the venue shopping behavior. The ACFs theoreti cal and analytical tools address the two limitations described above. includes a broad definition of policy actors involved in the policy process. definition of the subsystem includ es all potential venues within the subsystem boundaries Second, the policy subsystem allows the researcher to place boundaries on the research to avoid institutional features that could impact venue choices. The state level subsystem holds constant those institutional features commonly hypothesized in the venue shopping literature 5 Holyoke et al., (2012), for example, surveyed charter schools in three states boards. Another example is (2016) case study which focused on industry groups in Oregon. Others have examined a set of int erest groups within a subsystem. For example, Buffardi et al 2015) examination of nonprofits in Seattle, Beyers and ichigan, USA.

PAGE 33

20 as influential on multi venue choice (e.g., Constantelos, 2010; Holyoke, Brown, & Henig, 2012; Beyers & Kerremans, 2012), allowing for an empirical focus on policy actor attribut es. New York. This boundary contains the venue shopping inquiry within the state, and avoids interstate or international differences in lobbying rules or other rules th at may affect how often or why a policy actor shops a venue C venue can have major impacts in policy change (Nohrstedt, 2011), but the ACF neither describes nor explains the sovereign and another align with policy process (Sabatier, 1988). At the individual level, the ACF states decisions and information are shaped by their beliefs and perceptions. Therefore, Chapter 3 uses th e ACF as a starting point and then applies compatible theories within the ACF to develop hypotheses to such, venue shopping related ideas from the work of Schattschneider (1975), punc tuated equilibrium theory (Baumgartner & Jones, 1993) and other related scholarship (Pralle, 2004; Holyoke, Brown, & Henig, 2012; Constantelos, 2010) are applied to build hypotheses on the venue choices of policy actors. Chapter 3 focuses its inquiry into venue shopping through two common venue shopping question. It asks what factors influence the total number of venues shopped by a

PAGE 34

21 Chapter 4 and 5 Chapter 4 and Ch apter 5 examine how policy actors associated with local governments compare to policy actors associated with interest groups within a contentious policy debate. Policy process research on regional, state, and national environmental policy issues show local governmental representatives are active in policy advocates alongside other stakeholders and interest groups working to influence policy change (e.g., Sabatier, 1988; Sabatier & Jenkins Smith, 1994; Weible, 2006; Koontz et al., 2004; Blomquist, Schlager & Heikkila, 2004; Scholz & Stiftel, 2005). However, little is known about how local governmental representatives compare to other advocates within these broad policy debates. The ACF offers some insight into variation among coalition members who are affilia ted with level of analysis. Insights from the ACF regarding coalition beliefs, resources, political activities, and networks are applied to develop expectations related to how policy actors associated with local governments compare to policy actors associated with interest groups. C hapter 4 asks how the beliefs of local governmental policy actors compare to policy actors associated with interest groups. Chapter 4 b uild s on previous research that finds policy preferences of individuals are mediated by their organizational affiliation and on one of the usually advocate more moderate positions than their interest 106) Chapter 5 compares the resources, political activities, and collaborative networks of local governmental policy actors to policy actors associated with interest groups. To build the expectation that policy actors from different organizational types will have different resources Chapter 5 uses ideas from the resource based view of the firm

PAGE 35

22 (Penrose, 1959; Wenderfelt, 1984) and r esource dependence theory (Pfeffer & Salancik, 1978). To compare the political activities of policy actors affiliated with local governments with the resources of policy actor s affiliated with interest group allies Chapter 5 uses ideas from the p ublic m anagement literature to set the expectation that governmental actors will engage in different activities than non governmental actors (Rainey & Bozeman, 2000; Rosenbloom, 2015) Finally, to compare the networks of policy actors affiliated with local governments with the resources of policy actors affiliated with interest group allies Chapter 5 examines two aspects of policy actor networks. First, it explores the size of the netw orks. No expectations are developed for how the size of the network will compare across policy actors associated with local governments and interest groups. Second, it examines the network pattern of policy actors. Resource d ependence t heory is used to bui ld the expectation that the network pattern of policy actors associated with local governments will be similar to each other and different than the network patterns of policy actors associated with interest groups. Wrap up The four independent empirical c hapter s are presented next. The final chapter, Chapter 6, provides an over arching conclusion and discussion of the f uture direction of my research.

PAGE 36

23 CHAPTER II TRANSLATING BELIEFS : HOW POLICY ACTORS DEEP AND POLICY CORE BELIEFS AFFECT THEIR SECONDARY BELIEFS Introduction: Who Regulates A policy debate is a situation where individuals and/or organizations engage with government s to change or maintain the government s acknowledgement of and involvement in a problem. The result of a policy debate could include a redistribut ion of resources, new regulations, or other governmental tools to address a problem. The individuals or organizations involved in the debates use their limited resources to develop problem definitions, policy solutions, form coalitions, and engage with governing sovereigns. In contentious policy debates, there are typically two opposing groups who compet e to achieve disparate goals. Policy process scholars examine these policy debates and the choices made by those involved to develop theories that explain individual and group behavior and policy change. The Advocacy Coalition Framework (ACF) is one of the prominent policy process theory that focusses on group behavior within contenti ous policy debates. Through the ACF, individual and group behavior is explained, in part, through their beliefs. The ACF argues that a n individual s beliefs provide them motivation to act and that their policy goals reflect their beliefs. While policy scho lars have learned much about beliefs, a kind of belief that has not received much attention in policy process research is the preference over which level of government should regulate a policy issue Who regulates is a contentious question due to the na ture of regulations. Regulations are a specific type of policy tool used by governments to i) control the entry of a firm, its price setting decision, or its production levels, or ii) limit the impact of an economic

PAGE 37

24 activity on social or environmental welf are (Salamon, 2002, pg. 119). Regulations are solutions to problems identified as market failures (Weimer & Vining, 2011) Further, regulations are a coercive type of policy tool that governments use to exert their power to duce or secure goods (Salamon, 2002). R egulations not only influence the distribution of resources and actions of individuals or organizations, but also influence local, state, and national economies (Teske, 2004). Therefore, scholars consider regulations a critical area of public policy in the United States (Teske, 2004, p. 5) The decision over who regulates (i.e., which level of government regulates an issue) can be as contested and debated as what is regulated. In the United States, interest groups have fought long and fierce political battles to change which level of government has regulatory authority over a topic such as transportation, commerce, and social welfare (Teske, 2004). 6 Policy scholars identify two potential reasons th at policy actors attempt to change ideology or attitude toward government may affect his or her preference for who regulates. Looking back to the inception of the United States, the founders engaged in ideological debates as t hey outlined the federal system Their debates centered on where the federal powers would end and state powers begin (Middlekauf, 2007). These debates continue in modern day policy debates. P o litical ideologies seen in the Republican and Democratic parties are centered on the size and role of government. Donahue (1997) argues contemporary shifts in regulatory power were observed as the Republican Party under 6 State level regulatory authority seemed to dwindle after the Civil war and through the Great Depression as regulations were preempted by fe deral action. However, Teske (2004) notes that states maintained some power as they were required to implement many federal programs, and then, in the last part of the 20th century, devolution and deregulation at the federal level took its course, providin g even more power back to state governments and regulators.

PAGE 38

25 President Nixon, President Reagan, a nd the 1994 Republican Congress successfully fought to devolve federal regulatory authority to the states. However, ideology alone is not likely to completely explain preferences for federal or state level regulation. A s econd a policy actor may desire t o shift regulatory authority from one level of government to another for strategic reasons, based on the context of the debate. For example, as federal actors moved to shrink the federal government in the 1990 s other interest groups reacted (Donahue 1997; Teske, 2004). O rganized labor, environmental, and consumer groups moved toward open venues at the state level to pass policies to shore up changes or fill perceived voids in regulation at the federal level (Teske, 2004). Interest groups may also push for particular regulations at one level of government to preempt regulation from another (Teske, 2004; Hundley, 1986) Finally, policy actors may strategically seek open venues at multiple levels of government to push their policy agendas (Teske, 2004). This chapter tests how policy actor beliefs translate into a preference for a specific level of government to regulate an issue (referred to as preferred level of government) The A dvocacy Coalition Framework (ACF) is used as the theoretical foundation to address this question. The ACF model of the individual and its hierarchical belief system is applied to develop hypothesis on how beliefs affect their preferred level of government T he theory of venue shopping is also applied in this research to develop hypotheses for how context affects preferred level of government This research examines preferred level of government over a range of specific policy issues related to the topic of hydraulic fracturing based oil and gas development aka f racking. The research includes two localized issues and two broad issues (explained below) The variation in issue breadth is included because if an can be distinguished based on the level of

PAGE 39

26 abstraction, may also be a difference in how individual s conceptualize policy issue, based on This research uses empirical data gathered via an electronic a survey of policy actors involved in the state level f racking debates in Colorado and Tex as. The two s tate level fracking debates provide this research a rich testing environment for its question For example, within the two debates, t here are dif ferences between Colorado and Texas in which level of government regulates specific issues Further, the issues at the center of the fracking debates in both states vary in breadth (e.g., road damage a more localized issue vs. air pollution a broader issue ) Finally, the topic of who regulates is contentious. Within each state, policy conflic ts developed when policy actors attempt ed to change the level of government that regulates issues related to fracking. Contributions and M ap of the P aper The hypotheses are tested by building a model of government using multinomial regression with post estimation marginal effects The model will estimate the effect of government attitude and policy preference on preferred level of government for specific fracking related issue. This research adds to the ACF by empiricall y testing the relationship of beliefs within its three tiered hierarchical belief system. Specifically, it tests the relationship of broad, normative beliefs to specific preferences. Second, the research draw s in other theoretically backed factors that ma y affect a policy specific preferences Finally, this research adds to the ACF by a pplying the model of regulatory preference over a range of local ized and broad er issues The variation in issue breadth allows for the factors in the models to be co mpared and contrasted and provides

PAGE 40

27 scholars with more information on the nature of context when examining how normative beliefs translate into specific preferences. The following section builds the theoretical arguments through a review of the ACF and reg ulator choice. Next, the paper develops its hypotheses Then, the paper introduces the research setting, the fracking based oil and gas development subsystems of Texas and Colorado and regulatory structure. Next, the paper describes the methodology and results of the analysis. The paper ends with a discussion of the results, evaluation of the hypotheses, and research limitations. Theoretical Arguments and Hypotheses Development The Adv ocacy Coalition Framework (ACF) The ACF hierarchical belief system provides this research with a theoretical foundation to explore how broad beliefs may affect more specific preferences such as a red level of government to regulate an issue The ACF also acknowledges that contextual factors, such as the nature of the problem and existing rules, (Sabatier, 1988) Finally, the ACF is compatible with other policy process theories, such as the theory of venue shopping. Therefore, this research can apply insights from the venue shopping literature to examine how contextual factors affect the relationship between of higher level beliefs and lower level beliefs. In addition, and with respect to r provides a conceptual boundary around the fracking policy debates Policy subsystems include multiple policy actors engaging a multiplicity of policy making venues in policy debates over long periods of time. A policy subsystem is defined by a geographic area, a policy topic, and the actors within the geographical area involved in the substantive topic. In

PAGE 41

28 this research, the policy subsystem is defined by state boundaries, the topic of Fracking, and the p olicy actors involved in fracking related policy debates. T he ACF assumes subsystems are largely independent of each other due to the time and resources required for a policy actor to specialize on a topic to engage in a single subsystem (Heclo, 1978). Wit h this said, outputs from one subsystem can impact other subsystems and policy actors in different subsystems may interact (Zafonte & Sabatier, 1998) However, t his study, described in more detail below, assumes two independent subsystems for analysis of regulator preference These are the fracking policy subsystem in Colorado and the fracking policy subsystem in Texas. Hypothesis 1 H igher L evel B eliefs I nfluence L ower L evel B eliefs The ACF postulates that public policies are made through a process in which boundedly rational, belief motivated individuals form advocacy coalitions and engage governmental decision makers (Sabatier, 1988; Sabatier & Weible, 2007) A boundedly rational actor has cognitive limits and their beliefs act as a heuristic for pro cessing information (Scholz & Pinney, 1995). Further, an eliefs provide them with criteria by which they measure what is an appropriate policy response to a problem. 7 Therefore, beliefs are an important part of deriving range of preferences For example, the preference for what should be regulated, how it should be regulated, which policy making venue to select for a policy debate, and who should regulate the i ssue (Sabatier & Weible, 2007). beliefs using a three tiered hierarchical belief system. At the highest level of the hierarchy are deep core beliefs, then policy core, and 7 The ACF argues beliefs to be such a strong motivator that it allows individuals to overcome collection action problems, but, as Jenkins Smith, Nohrstedt, Weible, & Sabatier (2014) point out, thi s premise is under developed in the ACF, and not pursued here.

PAGE 42

29 finally secondary beliefs. In this hierarchy, beliefs range from the abstract to the specific (Sabatier & Jenkins Sm ith, 1993; Peffley & Hurwitz, 1985). The researcher can identify a of abstraction, how empirically based the belief is, the geographic scope of the belief with respect to the subsystem, and the level of difficulty to change the belief (Sabatier, 1988; deep core beliefs are analogous to basic world views and values. Deep core beliefs are considered constant and are not related to specific policy topics. Policy core beliefs are thought to be subsystem wide and define priorities such as whose welfare matters most in the subsystem, the role of government (including which level of government should regulate), problem identification and its seriousness at the subsystem level, and preferred policy solutions (Sabatier & Weible, 2007; Jenkins Smith, Nohrstedt, Weible, & Sabatier, 2014). Policy core beliefs are considered difficult to c hange but may shift over long periods of time of a decade or more (Sabatier, 1998). The lowest level beliefs are secondary beliefs. Secondary beliefs are not subsystem wide and typically associated with preference for a 2007 p. 196). Secondary beliefs are the most malleable of the three belief types, yet still resistant to change, and are more easily measured (Weible & Sabatier, 2006). Policy process scholars agree that the higher level, normative, deep core beliefs inform policy core and secondary beliefs (Jenkins Smith & Sabatier, 1994; Sabatier, 1998; Sabatier & Weible, 2007; Jenkins Smith, Nohrstedt, Weible, & Sabatier, 2014). ). For p core belief related to the relationship of humans and the natural environment will inform their policy preference on climate change, and potentially

PAGE 43

30 their secondary belief on addressing water use in their community. But, the degree to which deep core bel iefs constrain policy core and secondary beliefs remains unclear (Jenkins Smith & Sabatier, 1994; Weible, Sabatier, & Lubell, 2004). 8 Following the same example, ACF theory and empirical evidence from models of the belief system do not describe if an indiv preference for a policy tool to manage local water use (i.e. a secondary belief) To test the relationship between deep core and secondary beliefs this paper examines the eff ect of a specific deep core belief on a series of secondary beliefs. Recent scholarship finds individuals general attitudes from cultural cognition theory are consistent with the ACF (Ripberger, Gupta Silva, & Jenkins Smith, 2014; Jenkins Smith, Silva, Gupta & Ripberger, 2014). This research uses o ne of those general attitudes, the attitude toward government involvement in daily life as the deep core belief (Kahan et al., 2007; Gastil et al. 2016) and the pr eference for who should regulate as the secondary belief Hypothesis 1 is therefore : H1: Policy actors who believe governments should be involved less in daily life ( deep core ) will prefer lower levels of government to regulate issues (secondary belief). Policy actors who believe government s should be involved more in daily life will prefer higher levels of government to regulate issues Hypothesis 2 Context M ediates Lower L evel B eliefs The ACF acknowledges that the context surrounding the policy issue, such as the nature of the good, the current rules in place, and physical attributes related to the problem, 8 Researchers have also shown it is operationally difficult to differentiate between policy core and secondary beliefs (e.g., Olson, Olson, & Grawronski, 1999).

PAGE 44

31 constrain policy actors goals and action s (Sabatier, 1988; Jenkins Smith & Sabatier, 1994 Sabatier & Weible, 2007; Jenkins Smith, Nohrstedt Weible, & Sabatier, 2014 ). Another policy process theory, the theory of venue shopping, supports the idea that context mediates a policy actor policy decisions and goals. Venue shopping is defined as the s trategic act of a policy actor choosing a polic y making venue to hold a political debate. The theory of venue shopping postulates that within a policy debate with respect to the status quo (Pralle, 2003). For example, if the policy ac tor is dissatisfied with the status quo, they will engage in tactics to increase conflict (Baumgartner & Jones, 1993; Pralle, 2006; Schattschneider, 1975) One method a policy actor may use to increase the conflict and disrupt the status quo is to move the policy debate from one policy venue to another (Baumgartner & Jones, 1993). Conversely, individuals who desire to maintain the status quo may engage in tactics to contain a policy conflict, such as blocking another group or individual attempt to move th e debate to new venues (Pralle, 2006). While the theory of venue shopping focusses on strategic actions, it can help inform inquiry into the policy level of government to regulate a specific issue Indeed, Teske (2004) found that policy advocates pushed for regulators at the state level ov er regulators at federal level through a context informed strategic decision In other cases, advocates desired to preempt federal regulation. In other cases, advocates wanted to fill v oids in the governing structure left open by lack of federal regulation (Teske, 2004) If policy goals and strategic actions can be influenced by contextual factors, then secondary beliefs, such as regulator preference may be influenced as well. Hypothesi s 2 of this chapter

PAGE 45

32 Specifically, H ypothesis 2 that policy actors who desir e to expand political conflict will atte mpt to move a debate away from policy venues where the debates are currently held Conversely, policy actors who support the status quo will attempt to maintain the debates at the current policy venues (Pralle, 2006). Drawing upon the ACF and venue shopping literature produces an original hypothesis. Hypothesis 2 is: H2: Policy actors whose policy core beliefs do not align with the status quo are more likely to prefer regulators at levels of government that are differe nt than where they are currently administered. Lastly this paper expects the preference for which level of government to regulate an issue to change depending on the nature of the issue. The ACF identifies the nature of the problem at hand as a factor tha policy preferences (Jenkins Smith & Sabatier, 1994). Following Jenkins Smith and may be affecte d by the type of good in question and characteristics of the physical environment that relate to the policy issue (pg. 180). Therefore, I expect the type of issue in es with broader externalities, such as air quality are expected to be viewed differently than issues with localized externalities, such as road damage. To examine this expectation, the research includes issues with easily identifiable externalities (explai ned further in the operationalization section).

PAGE 46

33 Research Setting The topic of hydraulic fracturing based oil and gas development aka fracking, became a national issue following an expansion of the oil and gas industry that spurred from their technologica l advances in hydraulic fracturing and horizontal drilling. In the mid 2000s the oil and gas industry in the United States used these two technologies to expand its operations into known shale and tight sand formation as well as in previously unidentified shale deposits. Because of the success of these technologies, and the current price of oil, t he industry quickly found themselves in a modern day oil and gas boom. The industry expanded into areas accustomed to the industry, largely rural, extraction based communities, but they also grew into areas unfamiliar with oil and gas operations. For examp le, the industry began drilling near population centers schools and sensitive natural environments. While the industry expansion provided economic benefits, individuals across the country became concerned that fracking related development would negatively impact their health or the environment. Because of the risks associated with fracking multiple groups and communities mobilized to oppose hydraulic fracturing based development across the United States. Opponents of oil and gas development argue d that h ydraulic fracturing and the processes surrounding the extraction technique would negatively impact the environment and public health and safety (Food & Water Watch, 2015; Gallaher et al., 2014; Heikkila et al., 2014b; Pierce et al., 2013). Proponents argue d that not only were the environmental and health concerns unfounded, but the economic and national security benefits outweighed the risks (Hassett & Mathur, 2013; Heikkila et al ., 2014b; COGA, 2014 ; In the Matter of Changes to the Rules of the Oil & Gas C onservation Commission of the State of Colorado to Consider

PAGE 47

34 Hydraulic Fracturing Disclosure Rules, 2011 ). In response to advances and these debates, local, state, and federal lawmakers and agencies began updating or creat ing th eir oil and gas policies. The policies addressed a wide range of issues. For example, state agencies in Colorado and Texas updated their regulations in the mid to late 2000s to adjust where drilling could occur; to require industry disclose the chemicals u sed in fracking fluids ; the minimum distance between the wellhead and another building or land feature; and the amount of methane permissible into the atmosphere from a well ( Galbraith, 2012; Neslin, 2009) Indeed, fracking had become a contested policy i ssue in multiple states by the late 2000 s. In addition to the variety of fracking related issues debated by the opposing groups, there were also debat es over which level of government should regulate fracking. In the late 2000s and early 2010s e nvironmental and citizen interest groups were lobbying for more federal or municipal level regulations over the oil and gas industry For example, e nvironmental interest groups led multiple campaigns to develop federal level regulations for the disclosure of fracking chemicals and emissions f rom hydraulically fracked wells (FracAct 2011 and 2013) In 2012 the EPA released the first air standar d s for fracking wells and oil and gas pollutants that had yet to be regulated at the federal level ( EPA, 2012). 9 F urther, groups opposed to fracking sought to expand local regulatory authority in the late 2000s and early 2010s (Gallaher, 2015; Silverman, 2014 ) In response to the efforts by anti fracking groups to increase federal or local regulation over fracking, st ate regulators and pro fracking groups opposed changes to who regulated the industry A typical response by state officials 9 Federal level activity includ es debates in both congressional and regulatory venues (e.g., the Bureau of Land Management and the Environmental Protection Agency). Federal activity is not included in this research because the dataset used suggests that federal actors and issues have a played a minor role in state level debates (as of 2013) (Gallaher et al., 2014; Heikkila et al. 2014b; Pierce et al 2013).

PAGE 48

35 or industry groups to the local level oil and gas policy is a lawsuit against the municipality ( Gallaher, 2015; Sandberg, 2012 ; Fehl ing, 2015 ). Industry associations and other industry representatives often noted that the state was the preferred level of regulation (COGA, 2015) They argued that the industry was too nuanced to be regulated by the federal government (COGA, 2015) but local governments were not equipped to regulate the industry (Sandberg, 2012). Pro industry groups argue local level regulations would create a n unmanageable patchwork of regulation that would hurt the industry (UOGR, 2014; Haley, n.d.; COGA, 2016). The fracking related policy debates provide this research with a setting to examine interact. Within the debates, individuals represent a range of position s on whet her fracking should continue and are concerned about many problems related to fracking and oil and gas development that vary in scope ( Weible & Heikkila, 2016; H eikkila et al ., 2014b). For example, some individuals are concerned about dust and noise near a well, while others are concerned about fracking chemicals contaminating the water table (Heikkila et al 2014a). In addition, individuals have varying preferences over which level of government should regulate fracking (Heikkila et al ., 2014b). F inally, the current legal structure around oil and gas development provides this research with a regulatory landscape that is spread over multiple levels of government F or example, in the United States, state governments hold most regulatory authority over oil a nd gas development. However, within the two states used in this research, local governments have varying authority over nuisance issues (e.g., noise and dust) and the distance a well can be from neighboring buildings. These nuances are explained next.

PAGE 49

36 Two S imilar S ubsystems: Colorado and Texas Two state level policy subsystems, the fracking related oil and gas subsystems in Colorado and Texas, are used to examine regulatory preferences of statewide policy actors. Colorado and Texas were selected because of their strong similarities related to oil and gas development and regulation and similarities in the reaction to the rise in development due to new techniques of fracking and horizontal drilling. H owever, Colorado and Texas have different regulatory structures with respect to local level issues. In this sense, they are considered most similar cases (Gerring, 2007) and appropriate for hypothesis testing the affect current regulat ion has on a policy a issue. Regulatory S imilarities and D ifferences Both states have similar histories with oil and gas development and similar regulatory structures in terms of the state agency developing and promulgating ru les for development activities (STRONGER, 2011; STRONGER, 1993). Both Colorado and Texas have oil and gas activity dating back to the 1800s and recent booms attributed to horizontal drilling and hydraulic fracturing innovations and discoveries of shale dep state legislatures give authority to a state level regulatory body: The Railroad Commission of Texas (RCC) and the Colorado Oil and Gas Conservation Commission (COGCC), respectively. Both states have had multiple rule changes to update their regulations to accommodate the new technologies of hydraulic fracturing and horizontal drilling. They each have addressed disclosure of fracturing chemicals in 2011 (first in Texas and shortly followed in Colorado), but do address issues inde pendently and with slight variations (i.e., Texas has had more focus on water recycling and seismic activity and Colorado has

PAGE 50

37 examined the distance between wells and public or private structures an d water monitoring). o versee the rule change process and policy actors from industry, environmentalists, local representatives, and other organizations make comments during the rule making process. Regulation in Texas The RRC regulates a wide range of activities related to oi l and gas operations. At the time of this study rules and regulation for oil and gas regulation were contained in the Texas Administrative Code Part 1 Title 16 § 3. The RRC rules contain chapters regarding practice and procedure, informal complaint procedu re, oil and gas division, environmental protection, carbon dioxide, gas services division, pipeline safety regulations, LP gas (liquefied petroleum gas) safety rules, surface mining and reclamation division, coal mining regulations, regulations for compres sed natural gas, regulations for liquefied natural gas, alternative fuels research and education division, underground regulations for liquefied natural gas, alternative fuels research and education division, underground pipeline damage prevention, and adm inistration. Most notable is chapter three, which contains the regulations of the Oil and Gas Division. Chapter 3 section 1 through 107 cover rules ranging from water protection and other environmental management (§3.8, §3.22, §3.91, and §3.93), well desi gn; including casing, cementing, drilling, and completion requirements(§3.13), plugging and other completion activities (§3.14 §3.16), testing during and after well drilling (§3.17), safety and emergency management (§3.20, §3.21, and §3.84), hydraulic fra cturing chemical disclosure (§3.29), well spacing and density (§3.29, §3.38), waste (§3.98), fees, taxes and exemptions (§3.50, §3.78, §3.83, §3.101, §3.102, and §3.103), and penalties (§3.107).

PAGE 51

38 The Texas Commission on Environmental Quality is another key regulator of oil and gas operations. The TCEQ is the environmental agency for the State of Texas whose primary goals are to ensure clean air, clean water, and the safe management of waste The T CEQ and RRC share responsibility over waste, water quality, a nd injection wells. Since 1982, the TCEQ and RRC have used memorandums of understanding to clarify duties related oil and casing and/or groundwater protection recommen program as the Groundwater Advisory Unit (Railroad Commission of Texas 16 TAC Chapter 3 Oil and Gas Division, 2012). With respe ct to waste, and as of the time of this research, the TCEQ had responsibility over solid waste, which excludes waste resulting from oil and gas exploration, development, and production, and the RRC had jurisdiction over oil and gas waste. With respect to w ater quality, TCEQ sets water quality standards, and RRC enforces standards related to discharges and storm water resulting from oil and gas activities. TCEQ had jurisdiction over other water issues. For example, any surface water diverted for use in hydra ulic fracturing must obtain water rights through the TCEQ. Groundwater rights are obtained through courts and the State Legislature and managed either under the rule of capture, through individual land owners, or by Groundwater Conservation Districts (Hydr aulic Fracturing Frequently Asked Questions, http://www.rrc.state.tx.us/about/faqs/hydraulicfracturing.php). Regulation in Colorado. The Colorado Oil and Gas Conservation Commission is the regulating body in Colorado for all oil and gas and is housed in th e Department of Natural

PAGE 52

39 Resources. 10 Initially the mission of the COGCC was to promote the oil and gas industry and to prevent waste of oil and gas resources. Over time the mission was modified to include the protection of public health, safety, and welfa re and the environment (Pasternak, 1999). The permitting and drill site selection; the planning, site preparation, extraction, clean up, and surface recovery processes; en vironmental, health, and public safety requirements; and the required (consultations) or reactive (complaints and hearings) communication between the operator, land owner, mineral owner, downstream water users, the Colorado Department of Public Health and Environment, and the Colorado Division of Wildlife (COGCC Rule 201). In 2008, the COGCC completed a major overhaul of its oil and gas regulations to accommodate process changes related to and concerns of hydraulic fracturing and horizontal drilling. The C chemicals; Rule 317 Well casing and cementing; Cement bond logs; Rule 317B setbacks and precautions near surface waters and tributaries that are sources of public drinking water; Rule 341 monitor pressures during stimulation; Rule 608 Special requirements for CBM wells; Rules 903 & 904 pit permitting, lining, monitoring, & secondary containment; and Rule 906 requires Commission, CDPHE and the landowner of any spill that threatens to im pact any The Colorado Division of Water Resource s (DWR) is responsible for surface and groundwater use and consumption. Operators must maneuver through the water use program 10 In 1951, the Colorado General Assembly enacted the Oil and Gas Conservation Act and created the Colorado Oil and Gas Conservatio n Commission (COGCC) to carry out the provisions of the Act. The COGGC does not regulate exploration and extraction activity on Indian trust lands and minerals or the Southern Ute Indian tribe within the exterior boundaries of the Southern Indian Reservati on.

PAGE 53

40 administered by th e DWR to lease or purchase water rights. Under certain conditions, the Colorado Division of Wildlife (CDW) must be consulted in the development of plans for multiple or individual well location assessments. Further t he Colorado Department of Public Health and Environment (CDPHE) is involved in the permit to drill application process when a Local Government Designee (LGD) requests their participation, when the operator seeks a variance to specific rules related to the protection of public health, safety, we lfare, or the environment, or when the operator request s to increase well density (306.d.1.A.ii and 306.d.1.B). The Water Quality Control Division (WQCD) within the CDPHE is responsible for the permitting of discharges to surface waters (Stronger, 2011). T he COGCC has a with hydraulic fracturing (Stronger, 2011). Differences between Colorado and Texas While local regulatory issues appear similar, a major difference between the two states is in the role of local governments as regulators for local level issues. In Texas, state level regulators do not have jurisdiction over roads, leas es, pipeline easements, royalty payments, setback distances between the well and other buildings or natural features, or nuisance issues, such as traffic, noise, odors, (RCC website, n.d.). Rather, municipal governments in Texas have the authority over th ese issues. As such, many cities have developed or amended their ordinances regarding the exploration and production of oil and gas to include these issues (Barnett Shale Energy Education Council, n.d.). In Colorado, on the other hand, local authority is l imited to areas outside of drilling operations like road use and building permits. Nuisance issues such as noise, dust,

PAGE 54

41 body, the COGCC. The variation in local autho rity provides a setting to test hypotheses with In Colorado, debates over local and state regulatory control over oil and gas development date back to the 1980s and have led to multiple lawsuits between state and local governments. Local governments have placed moratoriums on drilling and attempted to use their land use and zoning authority to dictate where drilling could take place (Gallaher, 2015). State government representatives and the oil and gas in dustry argue the state has pre therefore loca l governments cannot intervene peaked in 2014 when interest groups for local regulations petitioned for multiple state level over the industry ( Richardson, 2014; Hostetter, 2014 ) At this time, Governor John Hickenlooper stepped in to the fray to negotiate a compromise A task force to investigate local control was created, and the ballot initiatives and current lawsuits were dropped as a result ( The State of Colorado, 2014 ). Texas has had some similar responses to local attempts to in crease their regulatory purview over development. However, the moratoriums set by Texas local governments have been short lived (Flower Mound) and, not until after this research was completed, and a ban on fracking related oil and gas development was set i n Denton, TX. This ban was quickly overturned by a state lawsuit ( Baker, 2015 ). Immediately following the overturned local ban, state officials attempted to cut off any future local action through by proposing a bill to make any attempt at local bans on hy draulic fracturing s illegal (Baker, 2015).

PAGE 55

42 Methods Population and Sampling A team of researchers, including the author, collected survey data used in this paper, as part of a larger project that encompassed Texas and Colorado. In this effort, we conducte d two sequential cross sectional surveys in Colorado and Texas, in 2012 and 2013, respectively. We target ed policy actors involved in the statewide oil and gas subsystems in Colorado and Texas. Policy actors, in contrast to the general population, defined as individuals who are professionally affiliated with an organization, involved in the policy area, and dedicate at least some time to influence, either directly or indirectly, the politics of the subsystem (Sabatier, 1988; Baumgartner & Le e ch, 2001). 11 We identified policy actors using a modified snowball sampling method. We began by identifying policy actors through internet searches of government documents, such as participant lists in fracking related rule making and legislative hearings Next, we expan ded the policy actor list by searching on line newspaper reports and documents published the policy actors we previously identified for additional names or organizations. Finally, we interviewed a subset the policy actors and asked them who should be inclu ded in the study. Through this process, we checked for biases in the search method by examining the organizational affiliation of the policy actors we identified. Our goal was to have a range of policy actors from the oil and gas industry, environmental gr oups, local, state, and federal governments, and the scientific community. We adjusted our search criteria to e nsure our final policy actor list represented those different organizational affiliations. T hese methods reduced the possibility that our population sample 11 An individual who submits an official comment on a policy debate, participates in a protest, or votes on a law related to a policy topic is not necessarily considered a policy actor. In the ACF, policy actors are differentiated from other citizens by the time they devote to an issue and the extent they specialize in the issue. Policy actors are differentiated from one another by attributes such as beliefs and resources and by behaviors.

PAGE 56

43 had coverage error, or the omission of key policy actors involved in hydraulic fracturing within each state (Singleton & Straits, 2010). Non probability sampling, such as this, is appropriate when there is not a pre made list, or other do cumentation, from which to sample or to create a sampling frame (Singleton & Straits, 2010). In total, we identified 398 policy actors in Colorado and 324 policy actors in Texas. Given the sample population size, we sent each policy actor an online survey (Singleton & Straits, 2010). Each survey was created and distributed through Qualtircs, an online survey tool. We gave each respondent three reminders to complete the survey after the initial request. We received a survey response from 142 of the 398 polic y actors in Colorado ( a 35.7% response rate ) and survey responses from 78 of the 324 policy actors in Texas (a 24% survey response rate ) The lower response rate in Texas limits the generalizability o f the results from this survey Variable Operationaliza tion The next section describes how I operationalize d the key concepts used to test this Secondary belief p referred level of government (Dependent Variable for H1 and H2). To measure secondary beliefs, I used a s urvey question that asked select only one level of government to regulate the following issues related to shale Response categories include o r egulation g overnment ; tate g overnment ederal g overnment 12 The issues included in this paper are public nuisance issues setback distances, air emissions monitoring, and water 12

PAGE 57

44 quality monitoring E ach answer is considered a secondary belief and was evaluated separately in a multi nominal regression model. of government should regulate a specific issue is considered a secondary belief because it is narrow in scope with respect to the subsystem, and more easily measured (Sabatier & Weible, 2007). I include four issues in this research that have distinguishably different externalities. The issues of w ater quality and air emissions are considered issues with broader externalities During the 2011 2012 chemical disclosure rule makings in both Colorado and Texas, p olicy actors who were against fracking argued that fracking fluids could contaminate the water table if a well casing failed, or surface waters if a spill occurred. With respect to air emission s policy actors as early as 2008 in Colorado were concerned that oil and gas operations were decreasing regional air quality through the release of volatile organic compounds (CDPHE, 2012 ; Dunn, 2013). The issues of setback distances and nuisance issues ar e considered to have more localized externalities. The nuisance issues were clarified in the survey to mean dust, noise, and light from the well site. Nuisance issues logically will only affect individuals who are near a specific well. Similarly, the issue of setbacks is considered a localized issue for the fact that the concern is if an accident occurs at a specific well, the local proximity is at risk for being damaged. While there may be differences in how one interprets the scope of these issues, I pres ent the results of the preference for level of and air issues are presented together and nuisance and setback issues are presented together. Deep core belief attitude toward government (Independent Variable for H1)

PAGE 58

45 To measure the respondent deep core belief about government, the survey asked two questions from cultural cognition theory about their general attitudes toward government involvement in daily life (Kahan et al., 2007; Gastil et al. 2016). overnment should put limits on the choices individual can make so they do not get in the way of what is good for society if th at means limiting the freedom and choices of individuals Respondents provided their answer on a four point Likert scale: 2 = Strongly Disagree, 1 = Moderately Disagree, 1 = Moderately Agree, 2 = Strongly Agree. These two questions measure levels of individualism vs collectivism (Kahan et al., 2007). Individualism and collectivism are two extremes on a scale measuring much governments shoul d intervene to aid in achieving cooperation (Gastil et al. 2016). This research uses a greement to these two questions to indicate that the respondent believes individuals should have less freedom of choice and governments should have a greater role in inf luencing individual decisions. Iterated Principal Component Factor analysis with Varimax rotation was used to governmental intervention. A negative score indica tes the respondent believes governments should have less decision making and a positive score indicates the respondent believes government should have more decision. 13 Policy Core Belief: Policy change pre ference (Independent Variable for H2) 13 Factor analysis resulted in a single factor, with eigenvalue 1.33 and each variable factor loading with a 0.877 after rotation. The factor ranged from 1.12 to 1.69. See Table 2 in Chapter 2 Appendix.

PAGE 59

46 Next, the survey asked about a general policy preference toward the fracking in current position in relation to unconventio nal shale development that uses hydraulic I kept respondents who stated that fracking should system I categorize the fracking as a policy core belief (Sabatier, 1988, pg 145, Table 1; Sabatier, 1998; Sabatier & Weible, 2007; Jenkins Smith, 1994; Jenkins Smith, Nohrstedt, Weible, & Sabatier, 2014; Weible, Sabatier, & McQ fracking aligns with the definition of policy core beliefs as it is: S ubsystem wide in scope ; h ighly salient as it is a fundamental policy position related to the substantive topic ; a t t he level of abstraction of this preference is low enough to be focused on a topic, but remains general in that it is referring to a broad range o f activities; and r elated to a policy that if changed, it would be considered a major policy change (i.e., if t he current policies that al low fracking were changed to stop fracking) 14 Opponents of oil and gas development viewed both the regulations and state regulators as pro development. Therefore, t he stop or limit group was considered to represent those policy actors who were against the status quo given that the national legal and 14 olicy preference. & Weible, 2007, pg. 195).

PAGE 60

47 regulatory framework allowed fracking and the practice had been expanding throughout the time of the survey. The continue at current rate group is the status quo group. T he status quo in the fracking debates v aries by state because of the variation in how states regulate fracking. Nuances in state regulatory structure s are described next. Control Variables Two control variables were used: the state and the organization type of the respondent. Designating the subsystem from which respondent came from, either Colorado or Texas, provide d this paper with the ability to compar e between issues that are current ly regulated at the state vs. the local government. At the time of this research, state level agencies were the primary developers and administrators of fracking related oil and gas development rules and regulations. However, Colorado and Texas differed on two local level issues: setback distances and nuisance issues. In Texas, municipalities controlled regulati ons over these issues. In Colorado, the state agencies controlled regulations over these issues. It is possible that other differences between the states could affect the strength of the arguments made using this variable. For example, institutional struc tures, such as legal standing may impact regulator preference (Jones, 2001). From the venue shopping literature, policy actors may choose the venue that has the legal standing, or legal ability, to make policy (Holyoke, Brown, & Henig, 2012) or the venue t hat is already active in addressing the issue (Mahoney & Baumgartner, 2009). Additionally, six organization types are included: environmental groups, oil and gas industry groups, federal government representatives, state government representatives, local g overnment representatives, and other. Other includes groups such as academics and consultants and the media.

PAGE 61

48 Model Selection and Analysis Techniques A multinomial logit model in STATA is used to test the relative influence of each variable on the regulatory preference of the respondents for specific issues associated with fracking. Marginal effects, a postestimation analysis, is then used to estimate t he overall effect of each independent variable, and interaction effects of independent variables, on The multinomial logit equation regulator is as follows: s preferred regulator m odel : Preferred level of government to regulate the issue i = Governmental attitude (H1) + Policy preference (H2) + State (Control) + Organizational type (Control). 15 Multinomial regression is an appropriate selection for modeling nonlinear systems with two or more categorical dependent variables (McFadden, 1999). Multinomial analysis is similar to a bivariate model (e.g., a logistic regression) in that the coefficients in the model outputs describe the effect on one outcome of the dependent variable with respect to another outcome. However, because multinomial models have more than two outcome choices, a base outcome option is selected by the user, and the coefficients for each independent variable represents a relative probability of selecting the output option of interest, with respect to the base outcome option. But the probability of one outcome is also impacted by the change in probability of the other outcomes. Th erefore, because of the number of equations involved in a multinomial regression, the interpretation is multi layered. B ecause of these the interrelationship of the coefficients and exponentiated coefficients in 15 STATA code: mlogit issue i c.gov_atti tude i.policy_preference i.state i.org_type, base(3) rrr. Variable names

PAGE 62

49 multinomial regression models, Rodriguez (20 Further, Williams (2012) notes the postestimation tool of marginal effects provide discussions beyond sign and statistical significance. A m arginal effects analysis is a postestimation tool for regression models and particularly helpful for interpreting the effects of independent varia bles in multinomial regression ( Williams, 2012; Rodriguez, 2017). Predictive margins, or marginal effects, provides the effect of the independent variable on the categorical outcome and whether the effect is statistically significant. 16 Marginal effects can be used to interpret continuous and categorical independent variables, but it should be noted that the interpretation of results is more straight forward for categorical independent variables (Williams, 2017a, 2017b; Royston, 2013). For categorical indepe ndent variables, the marginal effects analysis describes the discrete change in the predicted probability of a specific outcome, given the IV is = 1 and other variables are held constant. For continuous independent variables, the marginal effect is the ins tantaneous rate of change and is not always interpreted as the effect of a one unit increase of the IV on the DV (Williams, 2017a). Therefore, different marginal effects operations are considered in this research based on the nature of the independent vari ables for each hypothesis Marginal effects are displayed graphically to provide more intuitive information on the effect of each variable on the probability of a specific outcomes (Jann, 2013). 16 getting a predicted value of the dependent variable, and then averaging the predicted value (Becketti, 2003). This method is preferred for multin omial models as interpreting coefficients for multinomial models is difficult as they are discussed as relative probabilities (Rodrguez, 2017).

PAGE 63

50 Analysis and Results First, this paper present s the general results of the four full models and then describes the results related to H ypothesis 1 and then H ypothesis 2. Lastly, exploratory analysis with the control variables are presented. 17 The full mod els, presented below in Figure 3.1 and Figure 3. 2 show the marginal effects (dydx) of each variable on the predicted outcomes of for local, state, or federal regulator preference 18 Figure 3. 1 shows the marginal effects for the localized issue models: setback distanc es and nuisance issues. Figure 3. 2 sh ows the marginal effects the broader issue models: for air emissions and water quality. Each figure shows three columns one for each potential outcome of the dependent variable local government, state government, and federal government. All models show the marginal effects of each variable with 90% confidence intervals which is a visual indicat or for statistical significance at a p value of 0.90 or better 19 To interpret the categorical variables, use the first variable in each group of categorical var iables as the comparison variable. For example, the marginal effect stop/limit The marginal effect of the categorical variable is the change in probability that the resp ondent would choose the level of government when compared to the comparison value The marginal effect of the continuous independent variable, governmental attitude, is the instantaneous rate of change of the independent variable on the dependent variable. (Williams, 2017a). 17 n option, but only three respondents indicated they desired no regulation for the issue of setbacks, two respondents for the issues of nuisance issues and air emissions and only one respondent for the issue of water response was not included in the output options. 18 Model used Average Marginal Effects ( Williams 2011). 19 Appendix A, Chapter 2 Table 4provides the model results in tabular format with the exponentiated coefficients (relative risk ratios) and the exact p values.

PAGE 64

51 Note: 90% confidence levels shown. For each category of categorical variables, Policy Preference, State, and Organization Type, the first variable listed is the comparison variable, stop/limit, Colorado, and Environmental groups, respectively. See Appendix for full model tables with relative risk ratios. Figure 3. 1. Marginal effects (dydx) of Multinomial Regression Model: Setbacks and Nuisance Issues.

PAGE 65

52 Note: 90% confidence levels shown. For each category of categorical variables, Policy Preference, State, and Organization Type, the first variable listed is the comparison variable, stop/limit, Colorado, and Environmental groups, respectively. See Appendix for full model tables with relative risk ratios and exa ct p values. Figure 3. 2. Marginal effects (dydx) of Multinomial Regression Model: Air emission and Water quality.

PAGE 66

53 Overall R esults Governmental attitude For the issues of nuisance and setbacks, results show that government attitude has no effect on which level of government they prefer to regulate (Figure 3. 1) For the issues of air emissions and water quality, however, results show that respo a significant effect preference for loca l a nd federal regulators (Figure 3. 2) These difference s are examined in depth in the H ypothesis 1 section. Policy Preference For the issues of nuisance and setbacks, the results show that respondents who wish for fracking to continue or to expand are less likely to prefer local regulators and more likely to prefer state regulators, when compared to the stop/limit group (Figure 3.1) Those who wish for fracking to expand are less likely to prefer federal regulators when compared to the stop/limit group. 20 For air emissions and water quality issues the continue and expand groups are less likely to prefer local and federal regulators and more likely to prefer state regulators, when compared to the stop/limit group (Figure 3. 2) The differences are more pronounced and statistically significant for the issue of water quality. These differences are examined in depth in the H ypothesis 2 section. State For nuisance and setback issues respondents form Texas are statistically more likely to prefer local regulators and less likely to prefer state regulators for setbacks than those from Colorado. For air emissions and water quality respondents form Texas are statistically less likely to prefer state regulators and more likely to prefe r federal regulators 20 depends on the state, the comparison between status quo is only provided for H2.

PAGE 67

54 than Coloradoans. These relationships are considered further in below in the section on Hypothesis 2. Organization type For nuisance and setback issues, Oil and gas industry respondents are more likely to prefer state level regulator s and less likely to prefer local or federal level regulators than the environmental groups. The differences between governmental and other organization types with the base group vary between the issue of setbacks and nuisance issues. Statistically signifi cant differences are primarily found in the preference for local or federal level regulators. For air emissions and water quality, environmental groups generally less likely to prefer state regulators and more likely to prefer federal regulators than the o ther organization types. The only exception is federal government representatives are more likely to prefer federal regulators than the environmental groups for water quality issues.

PAGE 68

55 Hypothesis 1 Recall, Hypothesis 1 is: Policy actors who believ e governments should be involved less in daily life ( deep core belief ) will prefer lower levels of government to regulate issues (secondary belief). Policy actors who believe governments should be involved more in daily life will prefer higher levels of go vernment to regulate issues. To test Hypothesis 1, t attitude toward government on their preferred level of government to regulate an issue 21 Figure 3 shows the instantaneous rates of change of governmental attitude on the preference for local and federal regulator for air quality, water quality, nuisance issues, and setback distances (Williams 2017a & 2017b). 22 23 The marginal effect ( dydx ) value summarizes how a The independent variable of interest, government attitude, is continuous from 1.2 to 1.7. T he further the government value is from zero, the greater the effect of the covariate on the predicted outcome. If we assume a linear relationship throughout values of the governmental attitude we can interpret the results as such (Figure 3.3) F or the issue of air emissions, the results show that a one unit increase in govern mental attitude is associated with a 0.082 (8 percentage points) increase in the probability of preferring federal regulator. Therefore, for the issue of air emissions, respondent who prefer more governmental intervention are more likely to prefer federal regulators. Further, for the issue of air emissions, a one unit increase in governmental attitude, is associated with a 0.055 (5.5 percentage points) in a 21 See Appendix A, Chapter 2 Table 1 and Table 4 for more d issue and the tabular data on instantaneous rate of change of governmental attitude on preference for a specific regulator. 22 margin s, dydx (gov_attitude) 23 See Appendix A, Table 4 for the marginal effects of government attitude at representative values.

PAGE 69

56 likelihood to prefer local government regulators Both effects are statistically significant. For the issue of water quality, the results are similar to air quality. H owever, the effect of government attitude is only statistically significant on a likelihood to prefer local government regul ator s. For the small scale issues of nuisance issues and setbacks distances, t he results show that the effect of attitude toward government on their secondary belief their preference over which level of government shoul d regulat e is neglig ible and not statistically significant. Overall, t government on regulator pr eference varies by issue type. T he results partially align with the hypothesis that p olicy actors who believe governments should be involved less in daily life prefer lower levels of government to regulate oil and gas issues.

PAGE 70

57 P value <0.10, ** <0.05, ***, <0.01. Figure 3. 3. The instantaneous rate of change of governmental attitude on local and federal regulator preference for issues with broad and localized externalities.

PAGE 71

58 Hypothesis 2 Recall, Hypothesis 2 is: Policy actors whose policy core beliefs do not align with the status quo are more likely to prefer regulators at levels of government that are different than where they are currently administered To test Hypothesis 2, this paper examines the issues with localized and broad externalities separately. Recall that setbacks and nuisance issues were regulated at the state level o f government in Colorado and at the local level of government in Texas, and air and water quality issuers were regulated at the state in both Colorado and Texas. Therefore, the status quo is different for issues with localized externalities by state. For those from Texas, the expectation is respondents who are against the status quo those who desire fracking to be stopped/limited will be less likely to prefer local regulators than respondents who are for the status quo those who desire fracking to co ntinu e at its current rate or to expand. For those from Colorado, the expectation is respondents who are against the status quo will be less likely to prefer state regulators than respondents who are for the status quo. F or the issues in this paper with br oader externalities, there is no difference in who currently regulates between Colorado and Texas. Therefore, the expectation is individuals who are against the status quo will be less likely to prefer state regulators than individuals who are for the stat us quo F or each pair of issues this paper completed two analyses : First, the marginal effects of policy preference on the probability of regulator preference. Second, the marginal effects controlling for state.

PAGE 72

59 Marginal effects of policy preference for Issues with localized externalities: Setbacks and Nuisance issues. Figure 3. 4 shows the marginal effects of poli cy preference toward fracking on their preferred level of government to regulate the issues of setback s and nuisance s Eac h series in the figure represents a preferred level of government to regulate (local, state, or federal). The Y axis probability of selecting the level of government. T he X axis represents the policy preference categ ories (stop/limit, continue, expand) For the issue of setbacks, the results show respondents who desire fracking to be stopped or limited have a probability of 0.61 to prefer local regulators, 0.21 to prefer federal regulators, and 0.18 to prefer state r egulators. Conversely, individuals who desire fracking to continue at its current rate have a probability of 0.56 to prefer state regulators, 0.41 to prefer local regulators, and 0.04 to prefer federal regulators Finally, results in Figure 3. 4 show i ndivi duals who desire fracking to expand have a probability of 0.76 to prefer state regulators, 0.25 to prefer local regulators, below 0.00 to prefer federal regulators. The trends are similar for nuisance issues: as policy preference re lated to fracking moves from stop or limit, to continue, then to expand, the ir probability to prefer state regulators increases from 0.21 to 0.57. A notable difference is that individuals whose policy preference is for fracking to continue at the current r ate prefer local regulation (Pr(local regulation) = 0.53) more than state regulation (0.42). In for both local issues, the probability for preferring federal regulators across all groups does not go above 0.21 and moves toward 0 as policy preference change s from stop/limit to expand.

PAGE 73

60 Figure 3. 4 Marginal effect of policy preference on regulator preference for localized issues. 24 24 Margins over policy preference for different outcomes: margins, over(policy_preference) pr(out(2)) lev el(90); margins, over(policy_preference) pr(out(3)) level(90); margins, over(policy_preference) pr(out(4)) level(90).

PAGE 74

61 Marginal effects of policy preference between states for localized issues Figure 3 .5 preference for local, state, and federal regulators for setbacks and nuisance issues between those from Colorado and Texas. See Appendix A Table 5 12 for the differences between state marginal effects and their signif icance. For setbacks, the results show respondents from Texas are more likely to prefer local regulators than respondents for Colorado. However, respondents in both states who desire fracking to be stopped or limited are more likely to prefer local regulat ors than policy actors who desire fracking to continue or expand. Additionally, respondents from both states are more likely to prefer the state to regulate the issue if they desire fracking to continue or expand, than if the respondent desires fracking to be stopped/limited. While the results do show a significant difference between respondents from Texas and Colorado, the results do not support Hypothesis 2. For nuisance issues, there is no difference between respondents from Colorado and Texas in their p referred level of government for regulation. The results show similar desire fracking to be stopped/limited are more likely to prefer local regulators and less l ikely to prefer state regulates than respondents who desire fracking to continue or expand. While these results support Hypothesis 2 for respondents from Colorado, the general trends imply respondents who area against the status quo are generally against state level regulation and those who are for fracking are generally for state level regulation. Finally, given the ubiquitous low preference for federal regulators for local issues, and high preference for local regulators implies some directionality to th e regulator preference based on the nature of the issue, rather than which level of government currently regulates the issues

PAGE 75

62 Figure 3. 5 Marginal effect of policy preference, by state, on regulator preference for localized issues.

PAGE 76

63 Marginal effects of policy preference for i ssues with broad externalities: Air and water quality Figure 3.6 shows the marginal effects of policy preference on regulator preference. The preferences for who regulates are similar across both issues : respondents who desire fracking to be stopped or limited also prefer federal regulat ors over state or local regulators. Additionally, results show respondents who desire fracking to continue or expand also prefer state regulators over federal or local re gulators For example, for the issue of air emissions, the probability for preferring state regulator s moves from 0.29 if they desire fracking to stop or be limit ed to 0.70 if they desire fracking to continue and to 0.86 if they desire frack ing to expand Given that both issues are regulated at the state in Colorado and Texas, these results align with Hypothesis 2 Further, the relatively higher preference for federal regulators to local regulators for these shows a similar directionality in the preferred alternative regulator for issues with broad externalities as seen with the localized issues.

PAGE 77

64 Figure 3.6 Marginal effect on regulator preference for air and water quality over policy preference.

PAGE 78

65 Marginal effects of policy preference between states. Figure 3.7 shows the marginal er issues of air and water quality For both air and water quality issues, results show significant differences between re spondents from Colorado and Texas in the probability of preferring state and federal regulators and no difference between Colorado and Texas respondents on their preference for local level regulators. Given the regulations for air and water quality are bot h held at the state in Colorado and Texas, there is no theoretical expectation or explanation for this difference. Respondents from Colorado are more likely to prefer state regulators than local or federal regulators across a ll policy preference categories for both air and water quality issues. Even though the Coloradoan respondents who desire fracking to be stopped /limited are less likely to prefer state regulators than those who desire fracking to be continued or expanded, these results to do no t align with Hypothesis 2. Indeed, the shape of the curve between across the local, state, and federal marginal plots for the stop/limit group should be either U shaped or a diagonal line, showing either the probability to prefer local or federal regulator to be greater than the probability to prefer state regulators. Texan respondents however, show the hypothesized preference profile based on their policy core belief of whether fracking should be stopped/limited, continued, or expanded and the status quo of state level regulators. Texan respondents who wish fracking to be stopped/limited are most likely to prefer federal regulators and least likely to prefer local regulators for both air and water quality issues. Texan respondents who wish fracking to cont inue or expand are more likely to prefer state level regulators over local and federal regulators for the issue of water quality, and more likely to prefer state level regulators than local level regulators, and equally likely to prefer federal regulators for air quality issues.

PAGE 79

66 Overall, the results of the marginal effects analysis for the two broader and localized issues by state give mixed support for Hypothesis 2 that those who are against the status quo would be less likely to choose the level of government who currently regulates the issue The nuances observed in the marginal effects plots on preferred regulator between localized and broader issues, and those between states indicate the nature of the issue and the relationship between p ro and anti status quo to regulators is a p otential driver for preference.

PAGE 80

67 Figure 3.7 Marginal effect on regulator preference for air and water quality issues; state and policy preference interaction.

PAGE 81

68 Interaction of D eep C ore and P olicy C ore B eliefs on S econdary B eliefs The nuances identified in the results indicate anticipate the policy status quo correctly Re policy core belief their preference toward fracking affects their secondary belief their preference for which level of government should regulate, but it is not explained by which level of government currently regulates. he final analysis examines the interaction between a deep core belief (government attitude) and a policy core belief (preference related to fracking policy) on a secondary belief (regulator preference) for each issue. Figure 3.8 and Figure 3.9 s how the marginal effects of government attitude and policy preference on regulator preference for local issues. Figure 3.1 0 and Figure 3.11 show the marginal effects of government attitude and policy preference on regulator preference for broad issues. Fo r localized issues, the effect of deep core beliefs (indicated by the slope of each line) on the probability of preferring a specific regulator is insignificant. The driving factor for regulator preference is the policy core belief of policy preference. Fo r broad issues, deep core beliefs play a strong mediating role. For example, examining the issue of air emissions, the probability of an individual who desired fracking to be stopped to prefer local regulators is 0.4 when their government attitude score is 1.5 (desiring less government involvement) and nearly 0 when their government attitude score is 2 (desiring more government involvement). Similarly, the probability of an individual who desired fracking to be stopped to prefer federal regulators is ~0.22 when their government attitude score is 1.5 and over 0.70 when their government attitude score is 2

PAGE 82

69 Figure 3.8 Marginal effect of government attitude and policy preference on regulator preference for setbacks and nuisance issues. Figure 3.9 Mar ginal effect of government attitude and policy preference on regulator preference for setbacks and nuisance issues.

PAGE 83

70 Figure 3.1 0 Marginal effect of government attitude and policy preference on regulator preference for air emissions and water quality. Figure 3.1 1 Marginal effect of government attitude and policy preference on regulator preference for air emissions and water quality.

PAGE 84

71 Conclusion The a nalyses in this paper show no single variable explains preference for which level of gov ernment should regulate fracking development related issues. Overall, the analysis did not completely confirm e ither H ypothesis 1 or H ypothesis 2 (Table 3.2). While t he results of the broader issue models gave support mod els predicting regulatory preference for localized issues did not. However th e alignment of the results of the models predicting regulatory preference for broader issues with respect to Hypothesis 2 preference is likely due to of state, local, and federal regulators, rather than which regulatory body currently regulates specific issues. Indeed, in Colorado and Texas at the time of this research the state regulatory bodi es were supportive of oil and gas operations. This general view of state level regulators is seen in the results of each model and is consistent regardless of the post estimation effects. Figures 3.5 through 3.1 1 shows individuals are more likely to prefer state level regulators across all issues if they have the policy preference that fracking continue or expand. The figures also show individuals are less likely to prefer state level regulators if they have the policy preference for fracking to be stopped or limited. The r esults also show the policy core beliefs more consistently constrain secondary beliefs than deep core beliefs. deep core beliefs have no effect on regulator preference for localized issues but they have a significant influence on regulator preference when issues have broader externalities. The p olicy core belief on whether fracking should be stopped/limited, continued, or expand ed ha d the most significant effect on the r ir preferred level of government to regulate an issue Of the four issues examined in this research, t he stance on fracking significantly

PAGE 85

72 impacted their preferred level of governmental regulator However, res ults show the relationship of the core belief and the status quo as measured by which level of government currently regulates the issue is not supported. While results indicate respondents who desire fracking to be stopped or limite d also prefer local level regulators, the results also indicate they maintain this preference regardless of whether the issue is currently regulated by the state or local governments. Additionally, while the regulatory structure had the expected effect on preference for who regulates a localized issue of setbacks more preference for local regula tors it did not on nuisance issues. This may be that the Rule 800 in Colorado allows local governments to exempt themselves from the state rules related to nuisance issues but the rule does not allow them to create their own rules, like in Texas.

PAGE 86

73 T able 3.2. Results Summary. Hypotheses Localized Issues Broader Issues Hypothesis 1: Policy actors who believe governments should be involved less in daily life prefer lower levels of government to regulate oil and gas issues. Not Supported Supported Hypothesis 2: Policy actors whose beliefs do not align with the status quo are more likely to prefer regulators at levels of government that are different than where they are currently administered. Limited support Supported

PAGE 87

74 Discussion and Limitations based oil and gas development. Additionally, this research shows the aff ect that these factors nature of the issue in question While these findings fit well with the ACF (i.e., several beliefs and strategies shape the preferences of a policy ac tor) additional theory is needed to explain the The interaction effect of deep core and policy core beliefs on secondary beliefs observed in this research is worth exploring fu rther. Why, for example, is the effect of a deep core belief ( i.e., secondary belief ( i.e., preferred regulator) so strong for issues with broader externalities (Figure 3.9 ) but the effect of deep core beliefs evaporate s for issues with localized externalities (Figure 3. 10 ) ? Indeed, a dynamic in the ACF that needs further theory building manif est into policy goals on which the individual act s (Weible, Sabatier, & Lubell, 2004) In an early evaluation of the ACF, Jenkins abstract liefs continue to be hierarchical in nature. Weible, Sabatier, and Lubell (2004) provide empirical evidence that higher level beliefs inform lower level beliefs, but they found the line of influence between core and secondary beliefs may neither be direct nor include policy core beliefs. The results of this research provide additional empirical evidence for the linkage between higher and lower level beliefs but also the limits of deep core beliefs on secondary Focusing on the

PAGE 88

75 nature of the issue, such as the level of abstraction of the issue may offer some explanation. For example, air and water issues are broad and individuals may interpret those issues in a number of ways and so their deep core beliefs help them develop their preference for who regulate s. But the setback distances and nuisance issues are very tangible, so deep core beliefs are not needed as a heuristic in choosing their preference, and so their policy core belief has more influence influential in their preference of who should regulate. Another observ ation to consider further is why those who desire fracking to expand consistently prefer state regulators and those who prefer fracking to be stopped or limited prefer either local or federal regulators. One potential explanation for the relative preference for or against state regulators found in the political conflict and venue shopping literature is the presence of iron triangles and captured regulators. For example, across all four issues, respondents who desired fr acking to be stoppe d or limited were the least likely to prefer state r egulators and respondents who desired fracking to be expanded were most likely to prefer state regulators. This relative likelihood to prefer state regulators was consistent across the policy core belief related to fracking, regardless of the deep core belief (i.e. their attitude toward government) or if they were in the Colorado or Texas subsystem. Given that state regulators were considered by interest groups to be sympathetic to oil and ga s development, this may impact the preference for who regulates across all issues. A similar response is identified in the venue shopping literature: interest groups will seek out venues who are sympathetic to their cause (Pralle, 2003; Holyoke et al ., 201 2; Constantelos, 2010) and avoid venues where opponents are present (Hall & Daerdorff, 2006; Hojnacki & Kimball, 1998; Ley & Weber, 2015). However, this does not explain another dynamic on display in the analysis: those who wish for an expansion of frackin g have even more

PAGE 89

76 preference for state regulators than those who desire for continuation of fracking at its current that there is need to develop hypotheses rela ted to the direction of change, rather than simply change/no change. L eaving state preference aside a final observation of note is that the respondents preference for who regula tes appears to have some directionality (i.e., individuals prefer local regu lator s for the localized and federal regulator s for broader issues), indicating other significant factors exist. A more nuanced hypothesis is needed to address the directionality of regulator preference. For example, hypotheses that take the nature of the good, or issue in strategic level, such as the policy actor view on a problem. In this sense, the nature of the issue is driv ing which level of government is more efficient at its governance (North, 1984 ; North, 1990 ). As Buchanan and Tullock (1962) described the optimal size of government, they saw issues with larger externalities better handled by larger or higher levels of governments. Conversely, issues with smaller externalities are better handled by smaller governments. Buchanan and through observation of how metropolitan areas have found solutions to multi scale issues (V. Ostrom, Teibout, & Warren, 1961). V. Ostrom et al. (1961) argued a po lycentric, or multi level and overlapping, governance system is a necessity to manage events with a range of

PAGE 90

7 7 control does not match the boundary of the event they l ose their ability to regulate effectively (V. Ostrom et al., 1961, p. 835). Research on common pool resources provides empirical evidence of the importance of matching the governing boundaries to the resource and its users (E. Ostrom, 1990). Larger common pool resources are a challenge to manage with small scale, self governing appropriators because of a mismatch in governance and resource boundaries (E. Ostrom, 2005, 283). Therefore, a policy actor who is practically, or efficiently minded for solving pro blems may prefer local governments regulating smaller scale issues related to oil and gas development, and state or federal levels of government for regulating larger scale items. But then, their attitudes toward government or a specific regulator mediate that belief, driving some at extreme ends elsewhere. Given that this research only examined four specific issues, further research is needed in this area to confirm or deny the res ults. These alternative interpretations highlight some of the limitations of this study. For ses or building a stronger And findings. However, as with most limitations and unanswered questions, they also guide the direction of future research. As such, future work should include a broader range of issues with clear attributes, such as size of externality, incorporate theories on how prevailing institutional arrangements affe ct choice, and finally theories on how broader strategies of dominant and minority coalitions may play into their preference for who should regulate an issue. Indeed, one would expect different results in subsystems defined by a substantive topic that is n ot as contentious as fracking.

PAGE 91

78 CHAPTER III POLICY ACTOR S VENUE SHOPPING PATTERNS D FRACKING DEBATES Introduction : Venue Shopping No other strategic decision may be as critical for a policy actor during a contentious debate than where to debate the policy issue. Contentious policy debates involve multiple strategic decisions and actions made by policy actors (e. g. interest groups, governmental representatives, scientists, or the media). Strategic decisions include choosing with whom to ally, how to leverage focusing events, how to develop issue frames and narratives, which policy solutions to propose, and where to d ebate an issue. The decision of where to hold a policy debate, defined as venue shopping, is a specific tactic used by policy actors to expand or contain political conflict (Baumgartner & Jones, 1993; Schattschneider, 1975; Sabatier & Jenkins Smith, 1993; Pralle 2003). 25 Venue shopping also partially determines which governmental decision makers are active in addressing a policy issue, and to whom policy actors will advocate for their policy preferences. Indeed, policy actors select a venue, in part, based o n the decision makers associated with the venue (Holyoke, Brown, & Henig, 2012; Weber & Ley, 2015). Venue selection is a component of opening a new venue to a current policy issue, which is a mechanism for policy change (Kubler, 2001; Norhstedt, 2011). Fin inherent risk because in every policy debate, there is a winner, and there is a loser. 25 In this research, the term venue is limited to only governmental venues. In the broadest use of the word, venues could include media and other public outlets for policy debates.

PAGE 92

79 One reason policy actors venue shop is to expand political conflict. When a policy actor engages new set of governmental decision makers, they draw fresh attention to an issue (Baumgartner & Jones, 1991 ; Schattschneider, 1975 ). To affect such an expansion, policy actors may re frame a policy issue, find specific aspects of the is sue that can be addressed by a different venue, or choose a venue that is sympathetic to their point of view and problem definitions (Baumgartner & Jones, 1993). When policy actors find a sympathetic set of decision makers at a new venue, they can more eas ily break the established policy images or iron triangles (Baumgartner & Jones, 1993). Indeed, as part of a conflict expansion strategy, venue shoppers intentionally increase the attention paid to a policy issue. Once decision makers and the public turn th eir attention to an issue, a positive feedback loop, further increasing attention to the issue. Scholars show, increased attention is a necessary condition for major policy change (Baumgartner & Jones 1993; Schattschneider, 1975). Another reason policy ac tors venue shop is to contain conflict. Policy actors will select venues with decision makers at venues who are supportive of the status quo to contain a political conflict. They may achieve a similar effect by lobbying to keep debates within venues previo usly involved in the policy issue (Pralle, 2003). By maintaining the traditional venues involved in an issue, policy actors can minimize new attention to the issue. This maintains current policy images and can block the very policy change efforts described above (Pralle, 2003). Venue shopping is therefore both a possible strategy used by policy actors to either instigate or oppose policy change through what scholars describe as overcoming or engaging institutional friction (Baumgartner & Jones, 1993; Baumga rtner & Jones, 2005; Weible, Heikkila, deLeon, & study of multi level venue shopping shows, venue shopping can affect the very institutional

PAGE 93

80 arrangements that create friction or other path dependencies faced by policy actors engaged in contentious policy debates. Despite the effect venue shopping can have on policy debate outcomes, not every policy actor engages in venue shopping (Buffardi et al 2015). This is because it is a costly endeavor. Policy actors must acquire and use resources such as attention, time, money, and political connections and capital to engage a venue to advocate their policy position (Sabatier & Weible, 2007). Given t he cost and the potential impact venue shopping may have on the outcomes of contentious policy debates, policy scholars can learn about strategic policy actor behavior by ncing a reemergence with calls for, and development of, new theoretical models and quantitative models for such reasons (e.g., Ley & Weber, 2015; Constantelos, 2010; Holyoke et al 2012; Ley, 2016; Beyers & Kerremans, 2012). However, this research limits our understanding of policy actor strategic behavior in two ways. First, the above scholars do not include the range of policy actors involved in policy debates; rather they focus on a single interest group (e.g Holyoke et al., 2012; Ley, 2016; Buffardi et al., 2015; Beyers & Kerreman, 2012; Constantelos, 2010). 26 Second, the vertical or multi level venue shopping research designs are complex and include venues in multiple states or the nation as a whole, which draws theoretical focus away from the venue po licy actor relationship and onto institutional features affecting policy actor behavior (e.g., Beyers & Kerreman, 2010; Constantelos, 2010). Further, while the work is intended to examine how large institutional features affects 26 Holyoke et al. (2012 ), for example, surveyed charter schools in three stat focused on industry groups in Oregon. Buffardi et al. (2015) examined nonprofits in Seattle. Beyers and Constantelos (20 10) studied trade, business, or professional associations in Ontario, CA and Michigan, USA.

PAGE 94

81 the openness of venues, oth er institutional arrangements may exist (e.g., norms) that could affect political strategies like venue shopping decisions. Our understanding of policy actor behavior and venue selection can benefit from research that is designed isolate the inquiry more c losely policy actors and how their perception of the policy venue. This paper builds on the venue shopping literature by applying the Advocacy Coalition Framework (ACF) to examine the venue choices of policy actor. Not only does the ACF provide theoretica l guidance on key policy actor attributes affecting their actions, but its theoretical and analytical tools address the two limitations in venue shopping research. First, a dvocacy coalition includes all policy actors involved in venue shopping. Second, venue shopping research often examines patterns across state and national boundaries. This research This boundary simplifies the institutional features that could affect the study of venue choices, but still includes venues at multiple levels and branches of government (Sabatier, 1988). The research uses survey data from policy actors involved in New Y statewide fracking policy debates, and develops separate ordered logistic regression models to test two classic venue shopping questions 1. What factors influence the total number of venues shopped by a policy actor and; 2. tion of a policy making venue affect their shopping frequency at that venue? The remainder of the paper will first outline the theoretical underpinning of the research through a review of ACF. Then, the paper develops the hypotheses for each research

PAGE 95

82 quest ion using insights from venue shopping literature and the ACF. Following the hypotheses section, the paper describes the research setting subsystem. Next, the pape r operationalizes each variable and explains the regression m odels used for each research question. Next, the paper describes the results of the two models. The paper concludes with a discussion of the results and a reflection on the venue shopping literature. Theoretical Foundations The Advocacy Coalition Framework (ACF) Two constructs within the ACF are central to its theories of policy change and policy actor behavior in contentious contexts: the policy subsystem and the advocacy coalition. A policy subsystem is defined by a geographic area, a policy topic, and the policy actors within the geographical area involved in that topic. The policy subsystem construct simplifies analyses of complex policy process because it allows an analyst to define internal versus external influ ences on the policy subsystem and who qualifies as a policy actor. By focusing on a policy subsystem, scholars can identify and perhaps include multiple governmental decision making forums. This is an improvement on typical venue shopping research that loo ks at a targeted venue or limited set of venues. In this research, the subsystem boundary is drawn around a single state. Therefore, the analysis includes decision making venues at local and state levels of government and across each branch of government. This provides horizontal and vertical variation in venues and reduces the larger institutional features (e.g.,

PAGE 96

83 different state or national rules on lobbying) which could complicate the analysis on venue shopping decisions. 27 The ACF identifies a second con struct: the advocacy coalitions. These advocacy coalitions operate within subsystems. This construct of the ACF simplifies analyses at the subsystem level of analysis. The advocacy coalition construct recognizes the multitude of policy actors involved in p olicy change (e. g. interest groups, the scientific community, and the media, and individuals from all levels of government). However, it does not require the researcher to identify or examine each policy actor The ACF uses the boundedly rational individu al, and insights of group behavior from the policy network theory to inform the advocacy coalition concept (Sabatier, 1988; Sabatier & Weible, 2007). The ACF assumes that individuals use their beliefs as heuristics to simplify their understanding of new in formation and decision and policy goals, and help determine who is an ally or adversary within a policy debate (Sabatier, 1988). Further, to overcome individual physical and cognitive limit s, policy actors form coalitions to share resources and coordinate political activities to influence policy decisions. An advocacy coalition is therefore a broad network of policy actors with similar policy goals, who choose to act collectively to increase their ability to influence decision makers. With respect to venue selection, the ACF posits that advocacy coalitions strategically 27 The ACF does not provide guidance on differentiating between available venues. Sabatier (1988) discusses how multiple venues exist, but because debates move from one venue to another over time, using the subsystem as the unit of analysis captures this movement without getting in to the details. The importance of different venues and the activities around venue selection and engagement are acknowledged as important (Sabatier & Jenkins Smith, 1993 chapter 10).

PAGE 97

84 (Sabatier, 1988). Although the t advocacy coalitions, the implication is that policy actors in the coalitions aim to influence a set of decision makers who operate within one or more venues. The ACF neither describes nor explai another H owever contemporary ACF research has highlighted the importance of such a selection process (Nohrstedt, 2011). 28 Therefore, this research applies theories that are engage with governmental venues. Specifically, venue shopping related ideas from the Punctuated Equilibrium Theory (Baumgartner & Jones, 1993) and other related venue sho pping scholarship (Pralle, 2004; Holyoke, Brown, & Henig, 2012; Constantelos, 2010) are applied to build hypotheses on the venue choices of policy actors. In all, the ACF assists this research on the strategic activity of venue shopping by setting up an an alytical frame that includes all policy actors and defines the subsystem boundaries, but traditional venue shopping theories are needed to develop specific hypotheses related to its research questions. Hypotheses Development Research Question 1: What fact ors influence the total number of venues shopped by a policy actor? One way to achieve policy change is through political conflict expansion, which occurs when policy actors draw more people (i.e., decision makers, the public, interest groups, etc.) into t he debate (Schattschneider, 1975). One way policy actors achieve conflict strategically select decision makers who they believe will agree with their side of a polit ical 28 Nohrstedt (2011) found major policy change occurred after a new policy venue opened to the debate as a result of actions of the minority coalition members.

PAGE 98

85 conflict (Baumgartner & Jones, 1993 ). Once a set of decision maker s attention is on an issue, a positive feedback cycle of information and attention can be initiated. This facilitates the spread of the policy issue to other policy making venues (Baumgartner & Jones, 1993). A subset of venue shopping scholarship in US and Europe describes and explains differences in horizontal or vertical venue shopping activity. Horizontal activity includes movement between venues in different branches or domains of government within the same level of government (Holyoke, Brown, & Henig, 2012). Vertical or multi level venue shopping activity includes movement between venues in different levels of government (Princen & Kerremans, 2008; Beyers & Kerremans, 2012; Holyoke, Brown, & Henig, 2012; Constantelos, 1996, 2007, 2010). In research on v ertical and horizontal venue shopping, scholars have defined greater levels of shopping activity by the total count of venues shopped (Beyers & Kerremans, 2012) or number of levels contacted by an interest group (Constantelos, 2010). In short, scholars qua ntify the level shopping through counting the total venues shopped. As such, this research examines factors that influence the total venues shopped by a policy actor. Specifically, the research will include resources, policy actor type, and if the policy Resources. The ACF identifies a broad range of resources including finances, leadership, access to authority, access to scientific and technical information, and mobilizable supporters ( Sabatier & Weible, 2007; Weible 2007). The weight or relative importance of each of these influences is yet to be determined (Jenkins Smith, Nohrstedt, Weible, & Sabatier, 2014; Nohrstedt, 2011). However, the venue shopping literature theorizes that resour

PAGE 99

86 Constantelos, 2010; Holyoke et al 2010). Therefore, t he first hypothesis related to total venues shopped is: Hypothesis 1: Policy actors with more resources will shop more ve nues than those with fewer resources Although both theory and select studies show a positive relationship between a high resource level and increased venue shopping activity, this hypothesis is of continued interest because some empirical studies of venue shopping activity downplay the effect of available provided little explanatory power when examining interest group activity in multi level venue shopping settings (Constatnelos, 2010; Beyers & Kerremans, 2012). Policy actor type The tendency of a policy actor to engage in political activity may depend on their organization type. For example, scientists may produce information related to a policy issue and, therefore, be appropriately considered as policy actors within the subsyst em, but scientists may not engage with decision makers in the same way or frequency as interest groups. Likewise, governmental representatives at one level of government, or branch, may testify or advocate their preferences to another governmental decision making body, but have constraints on when or how they can act. Indeed, interest groups, by definition, advocate for policies to align with their positions (Baumgartner & Leech, 1998). Therefore, the expectation of this paper is that interest groups will h ave a greater amount of shopping activity than other policy actors identified in the policy subsystem. Hypothesis 2: Policy actors associated with interest groups will shop venues than policy actors associated with non interest groups.

PAGE 100

87 Beliefs of the po licy actors. Policy actors are belief motivated individuals, in that they seek to see their beliefs translated into policies through political activities, such as engagement with governmental decision makers (Sabatier, 1988). Therefore, if a policy actor d etermines the status quo policies do not align with their views, they are motivated to expend status quo policies, they may be less motivated to seek change. As venue shopping theory posits, an individual seeking change will attempt to expand the political conflict to facilitate policy change in their favor (Baumgartner & Jones, 1993; Pralle, 2006; Pralle, 2003). Therefore: Hypothesis 3: Policy actors whose beliefs de viate from the status quo will shop more venues than policy actors whose beliefs align with the status quo Control Variable: Organizational Focus. its venue selection (Constantelos, 2004; 2010). While Constantelos (2004 ; 2010) found interest groups shopping outside of their stated focus or mandated jurisdictions, the logic remains sound that an organization created to influence a particular level of policymaking would shop at that level more frequently than others would. to advocate a policy issue within a specific state within the United States, their organizational focus would be the state activity and communi ty building as their preferred mode to address a policy issue, then their organizational focus would be local. Depending on the context policy debate, organizations focused at one level of government may be more involved than others may. While no expectat ion is set in this research, the organizational focus is included as a control variable.

PAGE 101

88 making venue affect their shopping frequency at that venue? Another arm of venue shopping litera ture explains why a policy actor chooses one venue, or one type of venue, over another. For example, Buffardi, Pekkanen, & Smith (2015) identify three types of venues differentiated by the branch, the domain, and the level of government. 29 Others examine th e frequency that interest groups shop at a specific level of government (Constantelos, 2010), or the frequency of shopping within a set of venues of interest (Beyers & Kerremans, 2012), or the likelihood of shopping a venue (Holyoke et al., 2012). The seco among the venues available within a subsystem. provide a large knowledge base on policy ac tor venue selection. Theory and empirical studies selection. I categorize these into four groups. First, factors external to venue; second, factors directly associated with the venue or the decision makers in the venue; third, factors that describe the relationship between the venue shopper and the venue; and fourth, factors internal to the policy actor. Factors external to venue Pralle (2003) argues that whether th e venue has jurisdiction or a policy topic may impact a venue shopper s choice. Other scholars find the 29 While Buffardi, Pekkanen, and Smith (2015) provide three types of venues based on branch, domain, and level, one should note that these types may not be distinguishable if a data set does not include elected officials tive, City or County Council, Department of Neighborhoods, other city government unit, other state government unit, Republican 8). Each of these were legislative; (ii) domain type: bureaucracy, elected officials, political party; and (iii) level of government: local,

PAGE 102

89 may decrease the likelihood that policy engages that venue (Hall & Deardoff, 2006; Hojnacki 1998; Hojnacki & Kimball, 1998). Rel ated to the previous finding, recent scholarship argues that the relative strength between competing interest groups will affect venue selection (Ley, 2016; Ley & Weber, 2015). 30 Factors directly associated with the venue or the decision makers in the venu e. Holyoke, Brown, and Henig (2012) argue that the policy preferences of the decision makers within each venue, and how the policy actor views those preferences, will affect a policy and Ley (2015) argue policy actors will choose venues definition. Finally, policy actors are more likely to shop venues that are currently involved in their policy topic (Holyoke et al ., 2012; Mahoney & Baumgartn er, 2008). Factors that describe the relationship between the venue shopper and the venue. Policy actors are less likely to select a venue when the actor suspects the associated decision Hall & Daerdroff, 2006; Hojnacki, 1997; Hojnacki & Kimbal, 1998; Ley & Weber, 2015). On the other hand, policy actors are more likely to select venues when the policy actor expects the decision makers have policy preferences that align with the policy acto Holyoke, Brown, & Henig, 2012; Constantelos, 2010; Wright, 1992). Finally, policy actors are more likely to select venues that they believe to have more power or authority over the policy topic of interest to the polic y actor (Constantelos, 2010). 30 Interest group competition, as an explanatory factor, is more commonly examined when considering why an interest group becomes involved in an issue and which issues they choose to engage in (Baumgartner & Leech, 2001).

PAGE 103

90 Factors internal to the policy actor. affect which ve nue they select (Pralle, 2003; Constantelos, 2010). For example, an interest group with a state level mission will shop st ate level decision makers more than local or federal decision makers. A policy actor s resources and skill will also affect their venue shopping decisions (Holyoke, Brown & Henig, 2012; McKay, 2011; McQuide 2010; Ley and Weber, 2015; Ley, 2016; Pierson 2000; 2004). Some venues, such as courts take larger amounts of resources and specific skill sets, that may not be required at an administrative This paper focuses on the factors internal to the shopper and the relationship between shopper and venue. Specifically, the paper examines the effect of the perceived influence of decision makers on shopp ing propensity. Relative influence. A basic argument in the venue shopping literature is that venue selection is a strategic choice made by policy actors to maximize their use of limited resources for advocacy (Mazey & Richardson, 2001; Pralle, 2003, p. 2 49). One way policy actors reduce the risk of advocating at the wrong venue is choosing a venue they believe to have direct influence over their issue of interest. The influence of a venue may either be due to its jurisdiction or institutional claim to the policy topic (Jones, 2001; Pralle, 2003). A policy actor may also believe the venue has influence because the venue is already active in & Henig 2012; Mahoney & Baumgartner, 2009). In addition to a

PAGE 104

91 selection at a specific venue is not only based on how important they believe the venue is, but also ho w important the venue is when compared to the other available venues. Thus: Hypothesis 4: Policy actors are more likely to shop venues that they perceive to have more influence in the subsystem than venues they perceive to have less influence in the subs ystem. Relative agreement Each policy venue is prone to a particular policy image or images related to a policy topic (Baumgartner & Jones, 1991). A policy image is a set of beliefs and values that influences how the decision makers define a problem or a pproach solutions. In the context of policy change, policy actors can expand the policy conflict by taking their new policy ideas to a venue that is open to their policy image (Baumgartner & Jones, 1991). 31 Alternatively, policy actors can contain a politi cal by keeping policy debates at venues where the decision makers agree with the current policies. This research applies that policy actors are more likely to se lect a venue that has decision makers that the policy actor agrees with. Therefore, Hypothesis 5: Policy actors are more likely to shop venues with decision makers that the policy actor agrees with more, when compared to decision makers in other venues i n the subsystem. Control Variables: Resources. Just as resources enable policy actors to increase the given venue (Walker, 1983; Beyers & Kerremans, 2007). While no ex pectation is made for 31 The policy image arguments of PET driven policy. As noted above, venue.

PAGE 105

92 selection model. Control Variables: Other Venue Shopping Activity Policy actors may shop multiple venues within a subsystem. While there is exis ting research that suggests a policy (e.g., ecology of games (Lubell, Henry, & McCoy, 2010), this paper does not emphasize that work. However, this model of venue shopping this paper uses does include control variables to indicate where else the policy actor. Control Variable: Organizational focus (See above) Table 1 below summa rizes the paper s two research questions and the hypotheses associated with each research question.

PAGE 106

93 Table 1: RQ and Hypotheses Research Questions Hypotheses RQ1 : What factors influence the total number of venues shopped by a policy actor? Hypothesis 1: Policy actors with more resources will shop more venues than those with fewer resources Hypothesis 2: Policy actors associated with interest groups will shop venues than policy actors associated with non interest groups Hypothesis 3 Policy actors whose beliefs deviate from the status quo will shop more venues than policy actors whose beliefs align with the status quo RQ2 How does a policy making venue affect their shopping frequency at that venue ? Hypothesis 4: Policy actors are more likely to shop venues that they perceive to have more influence in the subsystem than venues they perceive to have less influence in the subsystem. Hypothesis 5: Policy actors are more likely to shop venues with decis ion makers that the policy actor agrees with more, when compared to decision makers in other venues in the subsystem.

PAGE 107

94 Research Setting fracking related policy debates occurring at the state, county, and municipal levels of government. Further, this research includes the policy actors involved in the poli cy debates related to fracking debates. These dimensions define the New York fracking policy subsystem (Sabatier, 1988). By drawing the subsystem boundaries in just such a way, this research has two empirical advantages over other quantitative models that examine shopping patterns across state or national boundaries. First, the state level subsystem holds constant those institutional features commonly hypothesized in the venue shopping literature as influential on multi ve nue choice (e.g., Constantelos, 20 10; Holyoke et al ., 2012; Beyers & Kerremans 2012). This nuance allows for an empirical focus on policy actor attributes and simpler quantitative models (e.g., ordered logit models vs. multi variate logit models with fixed and random effects in Holyoke et al. (2010)). Second, all policy actors are involved, allowing for an examination of shopping patterns of interest groups and policy actors, such as governmental officials and scientists, who are commonly left out of venue shopping studies (Holyoke et al., 2012; Ley, 2016; Buffardi et al., 2015; Beyers & Kerreman, 2012; Constantelos, 2010). 32 A third advantage of this research setting is that it was completed during the time of a de facto statewide moratorium on fracking. Because of this, day to day regulatory issues and rule makings related to oil and gas development had stalled and the political focus turned toward 32 Holyoke and colleagues (2012), for example, surveyed charter schools case study focused on industry groups in Oregon. Buffardi et al (2015) examined nonprofits in Seattle. Beyers and Con stantelos (2010) studied trade, business, or professional associations in Ontario, CA and Michigan, USA.

PAGE 108

95 the state level rule development for hydraulic fracturing and to what extent hydrau lic fracturing should be allowed within the state of New York. Indeed, policy actors had been active in affecting the outcome of the moratorium since its inception in 2008. In 2008, Governor David Paterson, in response to the technological advances in oil instructed the New York Department of Environmental Conservation (DEC) to update the general environmental impact statement (GEIS) to address new technologies for oil and gas extraction. Then during the updating process, New York state representatives voted to place a moratorium on fracking. While Governor Paterson ultimately vetoed the bill, he issued an executive order stating that no permits to drill would be issued until the supplemental general envi ronmental impact statement (SGEIS) was completed (Brown, 2011; NYDEC, 2011). In 2011, Governor Andrew Cuomo continued the moratorium (Brown, 2011). While the moratorium remained in effect until 2014, in June 2012 Governor Cuomo floated a plan to lift the high volume hydraulic fracturing ban in communities that support the techniques that are in the counties of Broome, Chemung, Chenango, Steuben and Tioga (Hakim, 2012). Also during this time, policy actors contested fracking policy in venues across multipl e branches and levels of government. For example, the DEC received over 260,000 public comments on the SGEIS drafts. Further, local governments across the southern regions of the state passed fracking bans and moratoriums (Meyer 2012; Brown, 2011). Accord ing to Fracktracker.org, between August 2011 and September 2013, the number of fracking moratoria and bans increased from 23 to 160 (Fracktracker website), indicating an active and growing grassroots anti fracking movement. In response, pro fracking groups were suing the local jurisdictions at local (Gottlieb, 2012; Meyer, 2012) and state courts

PAGE 109

96 (Brush, 2013). 33 The policy debates occurring at multiple levels of government, and across different branches of governments, give this research a variety of venue s hopping activity to examine. Methods A team of researchers, including the author, collected survey data used in this paper as part of a larger project that encompassed New York, Texas and Colorado. W e conducted a survey of policy actors in the New York fracking subsystem in the fall of 2013. Given that of a policy actor. We identified policy actors through internet searches of government documents, such as partic ipant lists in rule making and legislative hearings associated with the moratorium and fracking related oil and gas in New York. In total, we identified 379 individuals and sent them an electronic survey. After the initial invitation to participate, we sen t two follow up reminders. We received 129 individual surveys, a participation rate of 34%. The survey included questions used to operationalize most of the independent and both dependent variables for this paper I collected secondary data to identify and operationalize policy actor attributes, such as organization type and mission. The following 33 There are also inter state interests in the issue, such as the Delaware River Basin Commission. The Commission is an agreement between New York, Penn sylvania, New Jersey and Delaware and the Federal government which regulates the dispersion of water from Cannonsville, Neversink and Pepacton reservoirs to some decision making capability over whether or not to allow hydraulic fracturing to occur in their bounds, and banned natural gas drilling in 2009 (Erickson & Ansteotter, 2011). The Commission has since proposed revised regulations that would lift the moratorium and allow hydraulic fracturing in the basin (Erick & Ansteotter, 2011). However, the state authorities involved in the commission do not a proposed regulations have not been accepted and fracking in the basin is still not allowed (Darragh, 2012).

PAGE 110

97 section describes the operationalization of each variable, beginning with variables associated with RQ1and ending with variables associated with RQ2. Operationalizing Variables for RQ 1: What factors influence the total number of venues shopped by a policy actor? To operationalize the dependent, independent and control variables for Research Question 1, I used survey dat a and secondary data. Total Venues Shopped (Dependent Variable for H1, H2, and H3). To measure the dependent variable, total venues shopped by a policy actor, I used a survey question that asked respondents to identify the frequency at which they target e d a policymaking venue to achieve their political and policy goals related fracking The survey question asked respondents to identify the frequency that they shopped multiple venues in New York (Appendix A, Figure 1). I recoded their responses so that Nev of venues shopped for each respondent. The range of total venues shop ped was from zero to seven 34 Resources (Independent Variable for H1). I use a survey question with a battery of resources to develop the average resource capacity of the survey respondents. The survey asked the respondent to provide the capacity their organization had to mobilize financial resources for lobbying, f inancial resources for paying staff, support from members of the organization, support from the public or support from government officials. Their responses 34 Monthly and Weekly shopped and the magnitude of the effect of each coefficient, there was change in which coefficients were significant. See A ppendix B Table 9 for the full results of the al ternative model.

PAGE 111

98 were enumerated as follows: Not Applicable = 0, No Capacity = 0, Limited Capacity = 1, Moderate Ca pacity = 2, Substantial Capacity= 3. I calculated the average score over all resource types to create a continuous resource capacity score Organization Type (Independent Variable for H2) To measure organization type I collected secondary data to identify whether the organization the respondent represented was an interest group or not. I used Baumgartner & not only membership organizations bu t also advocacy organizations that do not accept members, businesses, and any other organization or institution that makes policy related & Le e ch, 1998, p. xxii). 35 If the organizations did not align with this definit ion, typically governmental, scientific, or consulting groups, I labeled them as non interest group. Beliefs (Independent Variable for H3). To measure beliefs, I used a survey question o your current position in relation to shale gas development that uses hydraulic fracturing. It should be: Stopped, combined the respondents who fracking and the industry was expand ing under the current conditions. One could argue that 35 Many respondents were legal representatives of various groups and their organization type was coded to match the organization the respondent represented.

PAGE 112

99 to continue. Organizational Focus ( Control Variable ) To determine the organizational focus of the respond level, state level, or national level advocacy. 36 If th e organization was not a traditional advocacy organization, such as an oil and gas operator, I location. For example, some oil companies drill exclusively in New York, while other companies drill nationally. I categorized these two types of actors with a state level focus and national level focus, respectively. Model for Research Question 1 To test the hypotheses related to the total venues shopped, I used an ordered logistic regression using the odds ratio operation. The equation is as follows: Model 1 : (DV) number of venues shopped = (H1) Average Resources of st Group Organization type + (H3) Belief + (C1) Organizational focus 36 Some difficulties with organizations such as Earthworks: a national organizatio n that works at the grassroots organizational focus, for example, was coded as a national level focus

PAGE 113

100 making venue affect their shopping frequency at that venue? To operationalize the dependent, indepe ndent and control variables for Research Question 2, What factors explain the propensity to venue shop specific venues available within a subsystem, I used three additional survey questions. Venue Shopping (Dependent Variable for H4 and H5). To measure ven ue shopping frequency I followed examples from the venue shopping literature (Baumgartner & Leech 1998, pg. 34; Holyoke et al., 2012; Buff ardi, Pekkanen & Smith, 2015). I used a survey question which asked, following organizations to achieve your political and policy goals related to shale gas development that 3. The survey included six venues: local courts, state courts, local government, state agencies, state legislature (either assembly or senate), and the models Relative influence (Independent var iable for H4) To measure the influence of a venue, I used a survey question, which asked the respondents to rate each venue. The question asked, Since 2008, how influential have the following organizations been in politics and policy about shale gas development that uses high volume hydraulic fracturing in 37 Next, I calculated relative influence by ranking each 37 Two state agencies and the two legislative cha mbers were included in the influence survey question. To align influence scores with the venue shopping dependent variables, each respective pair was combined by taking the

PAGE 114

101 respondent fluence score for each venue (i.e., Local Government, Local Court, State order. I used an ascending rank so that the venue with the highest perceived influence had the l = and all others as 1. By setting up the relative perceived influence varia bles so that the highest influence was the largest rank, I could align the direction of the variable with the direction of the hypothesis. Relative agreement (Independent Variable for H5) To measure a policy agreement with decision makers at a pol icy making venues I used a survey question that positions on shale gas development that uses high volume hydraulic fracturing in New included = with each venue. 38 Resources and Organizational focus ( Control Variables ). Same as above. used an ordered logistic model for each venue. The equation is as follows: score 38 Two state agencies and the two legislative chambers were included in the agreement survey question. To align agreement scores with the venue shopping dependent variables, each respective pair was combined by taking the ma e

PAGE 115

102 Model 2: (DV) Frequency of venue shopping at a specific venue = (H4) Relative agreement with the venue + (H5) Relative perceived influence of the venue + (C1) Average resource capacity + (C2) Shopping patterns at other venues + (C3) Organizational focus Analysis and Re sults Analysis RQ1: What factors influence the likelihood to increase the total venues shopped within a policy subsystem? Table 3 below provides the regression outputs for the total venues shopped. The outputs shown are odds ratios with Z statistics in parentheses and the level of significance indicated by the *s. The far left model is the base model. The base model included only a single dependent variable: the average resources of a policy actor. The full model, on the far right, included a dummy variable for whether the policy actor was associated with an interest group or not, dummy variables for the whether the policy act or was associated with a group with a local, state, or national focus, and dummy variables for the policy core belief of the policy actor. See Table s 1 through Table 5 in A ppendix B for full descriptive statistics related to research question 1.

PAGE 116

103 Table 3. Ordered logit results for total venues shopped, showing odds ratio. Total venues shopped: Resources only Total venues shopped: Org Type added Total venues shopped: Org Focus added Total venues shopped: Full Model Average Resources 3.670*** 3.739*** 3.370*** 3.316*** (5.06) (5.09) (4.46) (4.25) Non Interest group Comparison group for Interest Group Interest Group 1.335 1.917* 2.129* (0.81) (1.70) (1.94) Local Focus Comparison group for Organizational Focus State Focus 0.279*** 0.300*** ( 3.12) ( 2.89) National Focus 0.463 0.526 ( 1.25) ( 1.04) Belief: Stop/Limit fracking Comparison group for Position on Fracking Belief: Continue fracking at current rate 0.213** ( 2.24) Belief: Expand fracking 0.406** ( 2.27) Observations 108 108 108 107 Pseudo R square 0.0676 0.0693 0.0947 0.1168 Chi 2 p value 0.0000 0.0000 0.0000 0.0000 Odds ratios shown; Z statistic in parentheses; p<0.10, ** p<0.05, *** p<0.01.

PAGE 117

104 Discussion of Results of Analysis of RQ1 As one can see in Table 3, the Full Model overall explanatory power is limited (Pseudo R square = 0.1168). However, the model does provide information on the relative effects of policy actor resources, organizatio nal type, organizational focus, and the beliefs on the number of policy venues shopped within a subsystem. 39 With respect to Hypothesis 1 Policy actors with more resources will shop more venues than those with fewer resources the results provide suppor t. As seen in the Full Model there is a positive relationship between resources and the number of venues shopped. Further, for a one unit increase in average resources, the odds that the respondent shops the greatest number of venues (i.e., shopping 7 ven ues) versus the combined lower number of venues (i.e., shopping six five, four three two one or zero venues) is 3.316 times greater, given all other variables in the model are constant. This result aligns with expectations set by the ACF as well as th e wider venue shopping literature. Previous studies have offered mixed results. For example, Beyers and Kerreman (2012) found no positive impact of resources on multilevel shopping in Europe (measured in total venues shopped), compared to Holyoke et al. (2012) who found a positive relationship between resources and total venues shopped across three states in the U.S This result empirically supports the importance of resources as a deciding factor regarding whether a policy actor shops multiple venues. With respect to Hypothesis 2 Policy actors associated with interest groups will shop venues than policy actors as sociated with non interest groups the results also provide support. Interest groups venue shop more than non interest groups (e.g., governmental representatives, academics, or consultants). To interpret the categorical variables, the odds 39 The same model was run using simple OLS, considering the number of venues shopped as a continuous variable, with similar relationships betwee n the IVs and DV, and with an adjusted R square of 0.1634.

PAGE 118

105 ratio of the va riable is described with respect to the comparison group for that variable. Therefore, in the Full Model, respondents associated with interest groups are 2.129 times more likely to shop the greatest number of venues versus the combined lower count of venue s, given all other variables in the model held constant. These results support the general venue shopping literature. The results also justify the inclusion of non interest group policy actor shopping activity. In other words, interest group representative s are more likely to shop more broadly, but they are not the only policy actors engaged in venue shopping within a subsystem. Next, with respect to Hypothesis 3 Policy actors whose beliefs deviate from the status quo will shop more venues than policy ac tors whose beliefs align with the status quo the results offer support group. Results in the Full M odel column show the expand group, the odds of shopping a greater number of venues is 0.406 times lower than the stop o r limit group. This result aligns with the venue shopping literature. Given that at the national level the industry was expanding, I view the desire to 2013, during t he survey, the moratorium was in place and other research on this data shows very few survey respondents were happy with the moratorium (Heikkila et al ., 2014; Weible & Heikkila, 2016). Finally, the control variables show that policy actors from organizat ions with a state level focus are likely to shop fewer venues than those from organizations with a local level

PAGE 119

106 focus. However, there was no difference in the number of venues shopped between the respondents from locally focused organizations and those from nationally focused organizations. These results support the venue shopping literature. They also empirically support the reports of a mobilized anti fracking, grassroots movement in New York. Indeed, Mufson (2014) argues that local level anti fracking eff orts contributed to the eventual decision to ban fracking in New York, in 2014. making venue affect their shopping frequency at that venue ? Table 4 below provides the regression outputs for shopping frequency models for all six venues. A separate column shows the outputs for each venue. The outputs shown are odds ratios with Z statistics in parentheses and the level of significance indicated by the *s. Each venue shopping frequency model acco unts for between 32% and 64% of the overall variability in shopping frequency. Each model has an overall significance with a p value less than 0.0001 (Table 4). See Table s 6 in A ppendix B for full descriptive statistics related to research question 2.

PAGE 120

107 Ta ble 4. Ordered logit outputs explaining frequency of shopping at specific venues in New York. State Court Local Court Local Gov. State Legislature State Agency NY Governor's Office Rank of Influence of DV 1.121 10.65*** 1.272 0.923 0.924 1.273 (0.36) (3.14) (1.44) ( 0.34) ( 0.47) (1.55) Rank of Agree w/ DV 1.071 0.353** 1.275 1.272 0.717** 1.051 (0.26) ( 2.31) (1.56) (1.42) ( 2.09) (0.20) Average Resources 3.984** 1.484 1.442 1.012 0.877 1.334 (2.53) (0.59) (1.04) (0.03) ( 0.32) (0.71) shop state court 90.68*** 0.856 2.034 2.053 0.413 (4.36) ( 0.28) (1.12) (1.14) ( 1.41) shop local court 23.96*** 2.522* 0.55 1.065 2.211 (4.69) (1.72) ( 0.92) (0.10) (1.35) shop local government 1.284 3.309* 3.670*** 1.031 0.685 (0.54) (1.81) (3.41) (0.08) ( 1.10) shop state legislature 1.347 0.436 8.463*** 11.71*** 9.254*** (0.38) ( 0.71) (4.15) (4.16) (3.89) shop state agencies 2.562 4.119 1.259 8.444*** 9.289*** (1.18) (1.20) (0.48) (3.83) (4.12) shop NY Gov.'s office 0.394 1.12 0.577 7.705*** 7.718*** ( 1.39) 0.17 ( 1.32) 3.97 4.04 Local Focus Comparison Group for Organizational Focus State Focus 0.62 0.118* 0.361* 2.965* 1.269 0.379* ( 0.66) ( 1.87) ( 1.87) (1.66) (0.39) ( 1.68) National Focus 1.053 0.352 0.467 1.031 1.737 0.623 (0.06) ( 0.85) ( 1.11) (0.03) (0.61) ( 0.58) Observations 100 97 97 100 102 102 Pseudo R squared 0.5014 0.6423 0.3251 0.6177 0.5898 0.5368 chi2 p value 0.0000 0.0000 0.0000 0.0000 0.0000 0.0000 Odds Ratio, Z statistic in parentheses. p<0.10, ** p<0.05, *** p<0.01.

PAGE 121

108 Discussion of Results of Analysis of RQ2 With respect to Hypothesis 4 Policy actors are more likely to shop venues that they perceive to have more influence in the subsystem than venues they perceive to have less influence in the subsystem the models, overall, do not provide support. These results therefore run counter to the venue shopping literature and previous venue shopping frequency model results (Constantelos, 2010). T he perceived relative influence of the venue was only significant in the shopping frequency model of local courts. In the case of the local courts, for a one unit increase in the rank of relative influence, the odds of shopping more frequently increased by 10.5 times, given all other variables in the model are held constant. The reason for this could be that court cases are expensive and risky to poli cy actors, as the outcomes set legal precedent for future policy debates. The descriptive statistics of shopping frequencies (Appendix C) show local court venue shopping was the least common: only 20% of respondents shopped the local courts at a frequency of yearly or more frequent, compared to an average of 61% of respondents shopping at least yearly at all other venues ( Appendix B, Table 7) Further, local at al In addition, as of late survey local fracking bans and moratoriums in New York were a relatively new policy approach for those opposing fracking. For examp le according maps generated by Fracktracker.org, Figure A and Figure B, the number of municipal fracking moratoria and bans increased from 23 to 160 between August 2011 and September 2013 ( See Appendix B, Table 8 ).

PAGE 122

109 Figure A. Map of local level anti frac king actions, New York, 2013. Source: Karen Edelstein, www.fracktracker.org Figure B. Map of local level anti fracking actions, New York, 2013 Source: Karen Edelstein, www.fracktracker.org

PAGE 123

110 The use of local courts to challenge the local anti fracking decisions could have been lagging the municipal decisions as other policy actors developed a case and made the decision to challen ge the local bans or moratoriums in court. Further, in 2012, two of the first court cases challenging local bans were decided in favor of the local governments (Gottlieb, 2012). Therefore, taking local policy decisions to local courts was a new, risky, and arguably expensive proposition. Only those who felt the courts had the influence to change the anti fracking policy may have decided to engage this venue. While th e perceived influence of state courts did not correlate with higher shopping frequency, the amount of resources a respondent had was positively and significantly associated with shopping the courts. In comparison, there was no significan t relationship found between shopping and relative perceived influence for all other venues in the study The descriptive statistics in Appendix B, Table 7 show the venues with the highest perceived influence score, The New ( scored 1.6 and 1.4 respectively, where 2 is ), were also among the mo st frequently shopped venues. State agencies at least once per year. Respondents had similar shopping frequency at the state legislature and local government, but the average perceived influence for these venues were only 1.13 and 1.23, respectively. An alternate explanation could be that the non court state level venues were the established policy makers in NY for oil and gas development. For example, in 1963, New Yor k lawmakers enacted legislation to regulate oil and gas development (Stronger, 1994). In addition, the Department of Environmental Conservation (DEC), and the Division of Mineral Resources within the DEC, has the authority to promulgate rules related to oi l and gas development. The rules codified by regulators in the

PAGE 124

111 1970s remained in effect through the 1990s (Stronger, 1994). By 1992, state level regulations had even addressed a version of hydraulic fracturing (NYDEC, 2011). Therefore, it is likely that po licy actors involved in the fracking subsystem engaged with the state level venues, because state level venues controlled the substantive policies related to oil and gas. With respect to Hypothesis 5 Policy actors are more likely to shop venues with dec ision makers that the policy actor agrees with more, when compared to decision makers in other venues in the subsystem the models provide limited support. Results show a n two of the six venue shopping frequency models: the local courts and the state agencies. However, the agreement score and shopping frequency at these two venues correlated in the opposite direction than that hypothesized. Rather than survey respondents s hopping more frequently at venues where they agree with the decision makers, the results show they shop less frequently. As seen in the local court model in Table 4, the odds of a respondent shopping the local courts is 0.352 times lower for each unit incr agreement with the local courts. These results are counter to expectations set by the literature (Holyoke et al., 2010; Constantelos, 2010). One explanation for this result is that individual s go to court to argue against an action. In other words, only policy actors who wish reverse or cease a current decision or activity go to courts. Therefore, policy actors engage courts regardless of how much they agree with judges. If we look at the other significant variables in the local court model, we see policy actors who engaged the local courts were locally active, also engage in other courts, and believe that local courts are influential. These factors seem to explain involvement more, while the agreement with the venue is more of a descriptive condition.

PAGE 125

112 Further, as seen in the state agency model in Table 4, the odds of a respondent rank of relative agreement with the decision makers at the state agencies. Similarly while the relationship was not statistically significant, Appendix B Table 7 shows the lowest average One potential explanation for this unexpected outcome is that du ring the moratorium, policy actors were dissatisfied with the state agencies while the agencies developed new fracking related polices. This could have been the case especially for the DEC as it grappled with the new supplemental general environmental impa ct statement (SGEIS), offered the public proposals for review, and responded to hearings and public comments. Policy actors may engage with these venues because they disagreed with the policies that the agencies were proposing in order to affect change. P olicy scholars identified a similar phenomenon in Colorado during the 2011 2012 fracking chemical disclosure rule making process (Heikkila et al., 2014). During the chemical disclosure rule making process, both industry and environmental interest groups de veloped a negative view of the chemical disclosure rule by the Colorado Oil and Gas each group s feeling about the disclosure rule improved after the process was finalized (Heikkila et al., 2014). The negative views of policy actors may have been why they were advocating at the COGCC. At the time, the policy actors had an instantaneous disagreement with the status quo (i.e., the draft policy), and so were motivated to affect change in that policy document. As shown in the case of the state agencies and the courts show, further

PAGE 126

113 shopping at that venue. Turning attention to the control varia bles in Table 2, the average resources of the respondent were only significant in explaining the shopping frequency at the state courts: for a one unit increase in average resources, the odds of shopping the state courts more frequently increases by 3.98, given that all other variables are held constant. This could signal that state courts are expensive and only those with enough resources can make use of this option, regardless of whether the respondents believed the venue was influential, or whether or no t they agree with the venue. In the other five models, resources did not impact the frequency of shopping at specific venues, a result which aligns with previous empirical models examining shopping propensity at individual venues (Constantelos, 2010). The With respect to organizational focus results show survey respondents who were associated with organizations with a state level focus were 0.118 times less likely than respondents associated with a local focus to shop local courts, 0.361 times less likely to shop local ndents associated with organizations with state level focuses were 2.945 times more likely to shop the State Legislature than respondents associated with organizations with a local focus. These on impacts where they focus their advocacy resources (Constantelos, 2010). The set of variables displaying the most influence regarding shopping frequency in

PAGE 127

114 court shopping model in Table 4, if a respondent shopped at the local courts, the n they were 23.92 times more likely to shop the state court more frequently, given all other variables in the model are held constant. This indicates that those who shop one court system are more likely to shop the other court system. A connection such as this may be due to the court system appeals process. For example, Norse Energy Company took their case against the Town of Dryden, NY to the New York State of Court of Appeals after failing at lower courts (Brush, 2013). When all six models are examined t ogether, additional shopping patterns emerge. Not only are the respondents who shop one kind of court more likely to shop the other kind of court more frequently, but those who shop one state level venue also are more likely to shop other state level venue s more frequently. There is cross over shopping too. Respondents who shopped local governments also shopped local courts and sta te legislature more frequently. Given these venue shopping patterns and the mostly independent nature of venue shopping frequenc y to a respondent that venue, this research indicates there are other forces driving shopping choices. The relationships seen in shopping patterns could be the result of what Holyoke et al. (2012) relationships developed within policy subsystems that lend to the creation of an ecology of games (Lubell et al 2010). Coupling the idea that established at the state level non court venues and the idea that policy actors may develop a set of skills or resources to engage in a specific venue or set of similar venues (i.e. venues with similar the rules of engagement, information sharing, personal relationships, etc.) could

PAGE 128

115 explain both the results of the venue shopping frequency models in Table 4 and the distribution of shopping frequency seen in Appendix B Table 7. In this scenario, because m ost fracking related policy making in New York took place at stat level non court venues, individuals may acknowledge the influence of those venues, and, on average, shop at them most frequently, but the influence is not the driving force. Rather, the matc h up of the and hypothesis is listed and the outcome based on their corresp onding model.

PAGE 129

116 Table 5: RQs and Hypotheses. Research Questions Hypotheses Outcome RQ1 : What factors influence the total number of venues shopped by a policy actor? Hypothesis 1: Policy actors with more resources will shop more venues than those with fewer resources S upported Hypothesis 2: Policy actors associated with interest groups will shop venues than policy actors associated with non interest groups S upported Hypothesis 3 Policy actors whose beliefs deviate from the status quo will shop more venues than policy actors whose beliefs align with the status quo S upported RQ2 How does a perception of a policy making venue affect their shopping frequency at that venue ? Hypothesis 4: Policy actors are more likely to shop venues that they perceive to have more influence in the subsystem than venues they perceive to have less influence in the subsystem. Not supported Hypothesis 5: Policy actors are more likely to shop venues with decision makers that the policy actor agrees with more, when compared to decision makers in other venues in the subsystem. Not supported

PAGE 130

117 Conclusions and Limitations The Advocacy Coalition Framework is a proven tool for examining policy actor behavior in contentious politics. This research shows that analyzing venue shopping patterns through the lens of the ACF enables research on policy actor behavior related to the s pecific shopping literature in three actors than previously included in venue shopping models (e.g., Holyoke et al., 2012; Ley, 2016; Buffardi et al., 2015; Beyers & Kerreman, 2012; Constantelos, 2010) Indeed, the definitions (Baumgartner & Leech, 1998). The broad policy ac tor frame provided by the Second, the way this research drew the subsystem boundary at the state level eliminates institutional differences often present in other venue shopping models (e.g., Beye rs & Kerreman, 2010; Constantelos, 2010) while maintaining venues at multiple levels of government and branches of government. Third, as defined by the ACF, the subsystem includes all venues within a boundary Therefore, more venues are included in the stu dy than typical venue shopping research. While a strength of the research is that it included many venues within New York, however a limitation of this research is that the survey was not consistent in its question that asked respondents to indicate which venues they shopped. For example, the survey asked respondents to indicate the frequency they shopped categories of venues, like local governments, and it asked about specific venues, like the DEC and New York Department of

PAGE 131

118 Health. Therefore, frequency mod els may be more representative for some venue types and less representative for others. view of policy change, the individual shopping models include the perceived leve l of influence of a venue and perceived agreement with decision makers in policy making venues. Indeed, previous studies of this nature collected metrics to identify the alignment between policy actors and venues with respect to specific policy issues. Whi le policy actor perception of agreement should account for alignment, this study is obviously reliant on survey include venue level attributes or factors external ven ue. For example, venue level rules that strength of opponents at a venue. These design features may decrease the generalizability of the findings to venue shopping scen arios in multi national settings, or out venue level institutional features affect policy actor behavior. resources and beliefs to their political activities informs the ve nue shopping literature in two number of venues the policy actor shops (e.g., Beyers & Kerreman, 2012; Holyoke et al. 2012). The difference in the results between this paper and the other models could be in how resources were quantifie d, the types of resources included in the model, or difference in the research institutional features included in the different models. More research should

PAGE 132

119 shop m subsystem. These results support Baumgartner and & 2006) arguments that policy actors dissatisfied with the status quo will seek out new venues to who wanted fracking to continue or expand also shopped fewer ve nues. centric view and focus on policy actor beliefs was less applicable for explaining the frequency that a policy actor shops a venue. Indeed, in this s view of the venue, with few exceptions, did not explain their shopping frequency at that venue. However, this ffects their venue shopping decisions. interest groups follow their mission with respect to which decision makers they choose to lobby. Lastly, the results of the shopping patterns in one venue informs the shopping patterns in other venues. This finding suggests institutional arrangements (e.g. rules of engagement ), the skills needed to engage the venue and t he network associations of policy actors may be more influential in individual venue selection than previously thought. Previous venue shopping research Henig, 2012; Mc Kay, 2011; McQuide 2010; Ley & Weber, 2015; Ley, 2016; Pierson 2000;

PAGE 133

120 2004). Rules or skills required could increase costs to engage with specific venues or subsets of venues. S hopping patterns are of theoretical interest and an area ripe for future researc h (Lubell et al ., 2010; Holyoke et al ., 2012; Ley & Weber, 2015; Ley, 2016). To date empirical venue shopping models have failed to highlight the network relationships of policy actors. Indeed, Holyoke et al. ( 2012) noted this very limitation in their discussion of shopping propensity and a feature of their researcher design. A next step of this research is to examine the network like venue shopping relationships of policy actors within the subsystem and how venues may group along meaningful institution al characteristics beyond their vertical or horizontal location federalist system One way to do this is to combine the six individual models into a single multi variate model. A multi level multinomial model would improve on some of the limitations of the separate models in this research. For example, the total number of observations would increase and allow additional control variables. In addition, a multi level multinomial model could include a variable to quantify the network patterns that the six indi vidual models suggest.

PAGE 134

121 CHAPTER IV COMPARING THE BELIEF S OF LOCAL GOVERNMEN TAL POLICY ACTORS TO THEIR INTEREST GROUP ALLIES Introduction : Local Governments as Policy Actors Local governments are increasingly involved and influential in higher level policy making (Gamkhar & Pickerill, 2012). The most obvious way local government s influence higher level policy making is through their local policy decisions and programs that address policy issues of interest to higher level s of government Examples include local governmental policy making on the issues of climate change (Lutsey & Sperling, 2008; Feiock, Francis, & Kassekert, 2010), smoking laws (Shipan & Volden, 2006), and same sex marriage (Haider Markel, 2001). A strong body of research within the federalism literature describes how local government (e. g. municipal or county governments) decisions are made (Dahl, 1961), including explanation of internal decision making processes (Tiebout, 1956; Feiock, 2002; 2004) and external influence s on local governmental decisions (Berke & French, 1994; Frug & Baron, 2013; Peterson, 1981). In addition, policy scholars illustrate how local level decisions impact decisions in other jurisdiction (e.g., Shipan & Volden, 2006; G amk h ar & Pickerill, 2012) and how local level decisions heighten the broader policy debate (Riverstone Newell, 2013) Riverstone Newell (2012) argues the sway of local decisions are amplified when higher levels of government depend on local governments for service provision. Anothe r way local governments affect higher level policy is through their involvement in the policy process as a policy advocate. Policy process research on regional, state, and national environmental policy issues show s local governmental representatives are ac tive policy advocates alongside other stakeholders and interest groups working to influence

PAGE 135

122 policy change (e.g., Sabatier, 1988; Sabatier & Jenkins Smith, 1994; Weible, 2006; Koontz et al., 2004; Blomquist, Schlager & Heikkila, 2004; Scholz & Stiftel, 2005 ). However, less is known about how local governmental representatives compare to other advocates within these broad policy debates. This chapter applies th e Advocacy Coalition Framework (ACF) to unpack the policy advocacy role of local governments in debates that span beyond their jurisdiction and compare them to other advocates in the same policy debates Specifically, this chapter explores how policy actors associated with local governments compare to other groups of policy actors with respect to the ir beliefs This chapter uses i nsights from the ACF to build an expectation that policy actors associated with local governments will have more moderate beliefs than policy actors associated with interest groups, even when they belong to the same coalition (Jenkins Smith & Claire, 1993; Nohrstedt, 2005; Nohrstedt, 2010 ) The research question of how local governmental representatives compare to other policy actors in a statewide debate is explored in the context of hydraulic fracturing based oil subsystem from here on out) in 2013. Within the fracking subsystem, policy actors were divid ed into two opposing coalitions. One coalition was pro fracking and the other coal ition was anti fracking (Pierce, 2013; Heikkila et al ., 2014; Weible & Heikkila, 2016). The coalitions were made up of representatives from the oil and gas industry, mineral owners, environmental and citizen led anti fracking groups, the media, scientists, and municipal, county, and state government s (Weible & Heikkila, 2016; Heikkila et al ., 2013). Environmental groups oil and gas industry groups and local government s made up the three largest groups in the debates (Heikkila et al ., 2013). Further, l ocal governments

PAGE 136

123 representatives were present in both the pro and anti fracking coalitions (Weible & Heikkila, 2016; Heikkila et al ., 2013) examine how local governments compare to advocate g roups who they do and do not align with respect to their policy beliefs on fracking. The remainder of this paper outline s the theoretical backing of the research through a review of the ACF and then develop s hypotheses related to how local governmental po licy beliefs differ from policy actors associated with advocacy groups T he paper then describes in detail. Next, the paper operationalizes its key variables and then describes the analyses and results of how local go vernmental policy The paper concludes with a discussion of the results and a reflection on the contributions to the literature and the limitations Theoretical arguments and Hypothesis Development The Advocacy Coalition Framework (ACF) Since s by Paul Sabatier and his early work with Hank Jenkins describe the policy process throu gh the actions of advocacy coalitions in a policy subsystem. theory building and hypothesis testing is a set of attributes used to compare policy actors within the subsystem. This research uses one of those attributes, belief s, to compare local government representatives to other advocates within the Colorado fracking subsystem. Other c oncepts from the ACF that are relevant to this research include the policy subsystem, policy actors, and advocacy coalitions Each of these con cepts are described next.

PAGE 137

124 Policy Subsystem. The policy subsystem provides boundaries for ACF analyses. Sabatier (1988) recognized that the study of the policy process needed to move beyond the focus on single policy events and decision makers at a single venue. A policy subsystem enables this shift in many ways. First the subsystem view of the policy process acknowledges all policy actors engaged policy change related to a specific topic not just interest groups. For example, the ACF identifies scientist s, interest groups, news media, and decision makers at all levels of government as policy actors in the subsystem In addition to a policy topic and the policy actors engaged in policy debates on the policy topic, the policy subsystem is defined by a geogr aphic region (Sabatier, 1988). While a researcher studying the policy process could vary the geographic scope of the subsystem to change the level of granularity of their analysis of the processes within, researchers who apply the ACF typically choose to b ound the subsystem by national, regional, or state boundaries. Finally, w ithin the geographic region, multiple policy making venues are included. does not differentiate between the policy making venues within a subsystem. Policy A ctors and Advocacy Coalitions. Policy actors within a subsystem are individuals, usually professionally affiliated with an organization, involved in the policy area and dedicating at least some time to influencing either directly or indirectly the politics of the subsystem. In contrast, an individual who submits an official comment on a policy debate, participates in a protest, or votes on a law related to a policy topic is not necessarily considered a policy actor. The ACF differentiates p olicy actors with in a subsystem from other citizens by the time they devote to an issue and the extent they specialize in the issue. Within a policy subsystem, policy actors form one or more advocacy coalitions. An advocacy coalition is a group of policy actors within a p olicy subsystem who share common

PAGE 138

125 policy goals coordinate the use of their resources and collaborate on activities to achieve those policy goals. Additionally, and building on the model of the boundedly rational actor (Simon, 1957), the ACF theorizes that policy actors are motivated to form coalitions with other policy actors to overcome individual cognitive and physical limitations on information processing and resource utilization (Sabatier & Weible, 2007). Policy actors improve their chances of successfully influencing policy decisions when they act collectively in coalitions. The nature of interactions between advocacy coalitions within a subsystem ranges from cooperative to conflicting (Weible, 2008). In policy subsystems with a contentious substantive topic there are typically two or three conflicting coalitions (Weible, Sabatier, & McQueen, 2009), but can range between one and five (Weible, Sabatier, & McQueen, 2009). In contentious subsystems, the coal ition that maintains political control over policy decisions over extended periods of time is considered a dominant coalition and acts to keep the status quo or supports policy change that is congruent with their beliefs. When there is a dominant coalition the opposition, possibly mobilized in one or more the minority coalitions, seeks policy change to affect the status quo in ways that is congruent with their beliefs ( Sa batier & Jenkins Smith 1993; No h r stedt, 2010 ). Minority coalitions often seek allies f rom outside the subsystem or take policy debates to alternative venues than where debates are traditionally held in the subsystem (Fritschler 1983 ; Baumgartner & Jones, 1993; Browne 1990; Worsham 1997 ) Hypothesis Development Beliefs A key attribute of individual policy actor s and a foundational element of the ACF, is the belief system. The ACF categorizes belief s using a three tiered hierarchical belief system D eep core beliefs are at the base of the hierarchy, followed by

PAGE 139

126 policy core beliefs and secondary beliefs are at the top of the hierarchy D eep core beliefs D eep core beliefs are considered constant and are not related to specific policy topics. Policy core beliefs a re thought to be subsystem wide and define policy priorities such as whose welfare matters most in the subsystem, the role of government, problem identification and the seriousness of the problem at the subsystem level, and preferred policy solutions to th e problem (Sabatier & Weible, 2007; Jenkins Smith, Nohrstedt, Weible, & Sabatier, 2014). Policy core beliefs are considered difficult to change but may shift over long periods of time of a decade or more (Sabatier, 1998). Secondary beliefs are not subsyste m wide beliefs and are typically of the three belief types, yet still resi stant to change. Overall, individual beliefs guide problem perceptions and policy preferences, moderate information processing, and act as a cognitive heuristic to identify potential allies. Research about policy actor beliefs shows the policy preferences that divide and combine policy advocates into coalitions are based on their policy core beliefs, called policy core policy preferences, rather than on their secondary beliefs (Zafonte & Sabatier 1998; Sabatier & Weible, 2007). While actors coalesce into ad vocacy coalitions based on shared policy preferences, variation in policy preferences and goals exist between members of the same advocacy coalition. The v ariation in beliefs among coalition members has been attributed to both the differences in their indi vidual beliefs (Sabatier, 1988; Weible, 2006; Nohrstedt, 2010) and to Jenkins Smith & Claire, 1993; Nohrstedt, 2005; Nohrstedt, 2010) There are endogeneity issues with the organization affiliation argument

PAGE 140

127 (Sab atier, 1988), however there is evidence that an individual may have their beliefs or self preferences. For example, the policy preferences of governmental actors are influe nced by their interest for continued public support, and that this interest supersedes their policy core beliefs (Nohrstedt, 2005; 2010). 40 The hypotheses in this paper build on two ideas. The f irst idea is that policy preferences a re mediated by their organizational affiliation The s econd is one of the usually advocate more moderate positions than their interest p. 106) These two ideas are applied in this paper to test whether policy actors affiliated with local governments have distinguishably different policy preferences than their interest group allies. I argue that both local governments and interest groups r epresent their constituents, however the constituents of local governments have a wider range of values on a specific policy topic than the members of a specialized interest group aimed directly at that topic Therefore t o address th qu local governments compare to policy actors affiliated with other organizations with respect to 40 interest. The choice to focus on beliefs or self interest was a difficult one as Sabat ier developed the ACF. He wrestled with self interest and beliefs as two potential constructs to explain individual policy preferences and goals. Eventually he chose to focus on policy actor beliefs over their self interest because he felt beliefs were mor interest (Sabatier, 1988, p. 142). Furthermore, Sabatier recognized organization affiliation as a potential influence, via self interest, but also identified the endogeneity issue between organizational affiliation, s elf interest and beliefs (i.e., were views of an individual due to organizational interests, or did they choose an organization because their beliefs aligned with the mission of an organization). Despite his arguments to use beliefs to explain an individua scholars have focused on self interest as an additional individual level influence. Studies of policy actors in the Swedish nuclear subsystem found the policy preferences of policy actors from governmental organizations ar e led by their self interest for continued public support and that their self interest superseded policy core beliefs when formulating policy preferences (Nohrstedt, 2005; Nohrstedt, 2010).

PAGE 141

128 administrative agencies to be : H1: Local government policy actors will have more moderate beliefs than advocates affiliated with interest groups. And, with coalition membership considered the expectation is: H2: Within a coalition, local government policy actors will have more moderate beliefs than their interest group allies. Research Setting In Colorado, fracking related oil and gas development debates emerged in the early 2000s following the improvement and combination of horizontal drilling and hydraulic fracturing techniqu es. With these developments, and the discovery of new oil and gas deposits, oil and gas operations drew closer to urban populations not previously accustomed to the activity. At the same time, questions related to the environmental and public safety of fra cking began to question if the practice should continue. Those for fracking related development argued the economic and energy security benefits of development, while those ag ainst oil and gas regulatory body, the Colorado Oil and Gas Conservation Commission (COGCC) began receiving numerous complaints and concerns related to the new activ ity. In response, in 2007, the General Assembly passed a law requiring the COGCC promulgate new rules to reconsider impacts to the environment and public health, safety, and welfare (Neslin, 2008). Then, as the national attention turned toward the chemical s used in fracking (Fisk, 2013),

PAGE 142

129 disclosure debates, many of the same issues related to economic benefits and environmental and public health degradation were espo used (Heikkila et al., 2014). actors in the state level debates (Heikkila et al., 2013; Weible & Heikkila, 2016). When grouped by organizational affiliation, the majority of policy actors were either representatives from the oil and gas industry, environmental or citizen led groups, or local governments (Heikkil a et al., 2013). Local governments were actively engaged alongside other interest groups in the law and rule making processes (COGCC website showing county or municipal involvement since 1995; Committee on Oil and Gas meeting minutes from 1999; COGCC rulema king 2008, 2012, 2013). Further, these groups are divided along a normative position that either supports or opposes fracking related development (Pierce, 2013; Weible & Heikkila, 2016). With respect to local governments, there were some for, and some agai nst, fracking related development (Heikkila et al ., 2013). For example, the city of Longmont, Fort Collins, Lafayette, and Boulder County, each passed or considered moratoriums or bans related to oil and gas development (news articles for Greeley, Fort Col lins, Boulder, Lafayette; In 2012 Longmont and in 2013 the City of Fort Collins, Boulder, and Broomfield all approved moratoriums and the city of Lafayette made hydraulic fracturing illegal). In response, the Colorado Oil and Gas Association (COGA) and/or the COGCC have filed lawsuits against many of these local decisions. Local governments also worked with oil and gas operators to overcome issues and maintain production within their borders (e.g., using Memorandums of Understanding to create specific regul ations between an operator an d the municipality or county (Wilson,

PAGE 143

130 2012). For example, the town of Erie, CO spent created an MOU with two specific operators that had regulations that went beyond state law, but was limited in scope to the co signed operator s (Dunnahoe, 2013). fracking subsystem offers an opportunity to examine how local governments compare to interest groups with whom they do and do not align with respect to their beliefs on fracking. Methods Population and Sampling A t eam of researchers, including the author, collected data for this paper through a cross sectional on line survey of policy actors in the fracking related oil and gas development policy subsystem in Colorado in 2013. In this effort, we used a n on probabilit y sampling strategy because as a pre made list of policy actors did not exist f rom which to create a sampling frame (Singleton & Straits, 2010). Therefore, we used of a policy actor as the operational definition for sampling strategy We then employed a modified snowball to subsystem. We began t he modified snowball sampl e with internet searches and newspaper reviews to identify salient state level oil and gas develo pment policy debates related to fracking in Colorado From these debates we identified a seed list of policy actors based on the individuals and organizations who provided official comments or testimonies, or those listed as official stakeholders in state level debates. Next, we reviewed o nline newspaper reports and documents published by the policy actors in the seed list to identify other policy actors. Finally, we interviewed a subset of the identified policy actors and asked them who

PAGE 144

131 should be included in the study. To reduce risk of bias in our sampling methodology we iden tified the organizational affiliation of each p olicy actors and checked for over sampling within a given sector ( e. g. the oil and gas industry, anti fracking groups, governmental affiliates, scientists, etc.). We adjusted our s earch criteria to ensure a r epresentative sample from the different organizational affiliations. The methods we sued reduced the possibility of coverage error (Singleton & Straits, 2010). Our efforts identified a list of 398 policy actors in We sent eac h an on line survey: 142 policy actors responded to the survey a response rate of 35.7%. The survey included questions used to operationalize the independent and dependent variables for this paper Operationalization Organizational groups (IV for H1 and H2). I used a survey question to identify the The survey asked each respondent to select organization types they most closely represented. In the analysis, I combined the oil and gas industry group I combined into the environmental group I combined into an group News media survey recipients did not respond to the survey and were removed from the study. Finally, I kept l ocal, state, and federal government respondents in separate groups. 41 41 The local government group included municipal, county, water dis trict, and local governmental association groups. Only four of thirty nine local government respondents represented local governmental associations. See Appendix C Table 1 for breakdown of respondents by organization affiliation.

PAGE 145

132 Policy Preference: Policy Core Belief and Coalition membe rship (DV for H1 and H2) To measure the first policy core belief, I used a single survey question which asked respondents to natural gas development that uses hydraulic fracturing. It should be... Stopped ; Limited ; C ontinued at the current rate ; Expanded moderately ; or Expanded extensively positio n & Weible, 2007, pg. 195). For the Hypothesis 1, the policy position is scored based on st op = 1, limit = 2, continue at current rate = 3, expand moderately = 4, and expand extensively = 5. F or H ypothesis 2 I transformed the policy position response into a dichotomous variable where stop or limit = 0 and continue, expand moderately, and expan d extensively = 1. I follow previous research by Pierce (2013) who identified two coalitions that aligned with this division. I establish policy actors are anti fracking if they stated hydraulic fracturin g should be stopped or limited and a re pro fracking if they stated hydraulic fracturing should be continued at its current rate, expanded moderately, or expanded extensively 42 Problem Perception additional policy core beliefs (DV for H2) Th e second used in this paper to compare respondents associated with local governments to respondents associated with advocacy groups is their 42 tilized the same data set used in this study. Pierce used 12 problem perceptions policy subsystem. Pierce (2013) estimated coalition membershi p using a cluster analysis of Manhattan distances preference score and a composite of 12 problem perception scores. His analysis found two coalitions: an anti fracking and pro fracking coalition. I only use the policy preference question to identify coalitions for two reasons. First this paper uses twenty problem perception questions, rather than the 12 problem perception 2013) coalition analysis. These eight additional questions have potential to be viewed differently by local governments than other group affiliations and may skew the Manhattan distance s, only two were not placed into the pro fracking or anti fracking coalition that aligned with their policy preference score.

PAGE 146

133 problem perceptions related to the substantive topic of the subsystem (Jenkins Smith & Sabatier, 1994). The survey asked re spondents to Please indicate the extent to which the following issues are current problems related to natural gas development that uses hydraulic fracturing for 20 issues. The respondents indicate d the severity of the problem on a 5 point Likert scale fr om not a problem = 1 minor problem = 2 moderate problem = 3 serious problem = 4 to severe problem = 5 The survey included issues related to environmental impacts of fracking based oil and gas development, the state of current regulatory structure, reg perception score for the specific issue.

PAGE 147

134 Analysis and Results Hypothesis 1: Local government policy actors will have more moderate beliefs than advocates affiliated with interest groups. This paper uses multiple methods t o compare the beliefs of local governmental policy actors to policy actors associated with other advocate group s. First, an Analysis of Variance (ANOVA) with a post hoc Fisher Hayter pairwise comparison policy position score is used to determine which groups o f respondents have statistically different policy core beliefs Results in Table 4. 2 below show the mean policy position score s of respondents associated with local governments are statistically different than the mean policy position score s of respondents associated with environmental and oil and gas industry groups. Further, the results show that the policy position score s of local governmental respondents (mean of 3.08) are more moderate than the policy position scores of respondents associated with environmental (mean 1.64) or oil and gas industry groups (mean 3.87) These results support Hypothesis 1 43 43 Within the local government category, four respondents were associated with local government association groups. Of these four respondents, two held the position that fracking should continue and two held the position that fracking should be limited. These positions are within the average positions of respondents associated with a single local government entity (i.e. a municipali ty or county). See Appendix C Table 1 for descriptive statistics for each policy actor group by their policy preference score and Appendix C Table 2 for ANOVA results and Appendix C Table 3 Fisher Hayter pairwise comparison results.

PAGE 148

135 Table 4. 2 Pairwise comparison of fracking policy position scores using Fisher Hayter method. Group 1 (mean score) vs Group 2 (mean score) Difference FH Test Local Gov (3.08) vs Oil and Gas (3.87) 0.79 6.05* Local Gov (3.08) vs Env ironmental g roups (1.64) 1.43 9.98* Local Gov (3.08) vs Fed Gov (2.77) 0.31 1.66 Local Gov (3.08) vs State Gov (3.18) 0.10 0.53 Local Gov (3.08) vs Academics (3.33) 0.26 1.20 Note: stop = 1, limit = 2, continue at current rate = 3, expand moderately = 4, and expand extensively = 5. The complete results of the Fisher Hayter pairwise analysis is found in Appendix C Table 3

PAGE 149

136 Next, the paper uses correspondence analysis (CA) to compare the fracking policy positions of respondent s when grouped by their organizational affiliation Figure 1 below shows the results of the CA. In Figure 1, the position is denoted by the dots and th eir organization al affiliation is denoted by the triangles The total inertia of the model is 0.71, meaning 71% of the variation in policy position is accounted for by the model. Ninety six percent of the is explained along two dimensions. The results show respondents with more extreme positions are associated with interest group s and respondents with moderate positions are associated with governments. For example, respondents associated with environmental groups cluster around the policy positi on to around the policy position See Appendix C Table 4 Finally, respondents associated with governments cluster a policy position The CA results corroborate the Fisher Hayter pairwise analysis and visualizes the similarities and differences in policy core beliefs between the local governments and advocacy groups (Figure 1). These analyses support H1 that policy actors associated with local governments will have more moderate beliefs than those from interest groups

PAGE 150

137 Figure 1. Correspondence Analysis of Policy Position and Organization affiliation.

PAGE 151

138 Hypothesis 2: Within a coalition, local government policy actors will have more moderate beliefs than their interest group allies. To test Hypothesis 2, respondents affiliated with local government s were separated into pro fracking and anti fracking coalitions us ing the methods described above using their policy position response to fracking. The local governmental r espond ents who stated they desired fracking to be stopped or limited were placed in the anti fracking coalition, while those who responded that they desired fracking to continue at its current rate or to expand were placed in the pro fracking coalition. Table 3 below shows the number of respondents from each organizational affiliation who desire fracking to be stopped/limited or continued/expanded. Twelve of t he thirty nine local government respondents indicated they desired fracking to be stopped or limited, placing them into the anti fracking coalition, and aligned with environmental respondents. Twenty seven of the thirty nine local government respondents indicated they desired fracking to be continued or expended, placing them into the pro fracking coalition, and aligned with the oil and gas industry respondents. 44 44 One individual from individual benefits. State and Federal government, and academic respondents were split to varying degrees into the two coalitions.

PAGE 152

139 Table 3. Coalitions based on policy position Fracking p olicy position Organization Affiliation Stop/Limit Continue/Expand Total Federal Government 5 8 13 State Government 1 10 11 Local Government 12 27 39 Academics and consultants 2 7 9 Oil and g as groups 1 38 39 Environmental g roups 28 0 28

PAGE 153

140 With the local governmental respondents divided into pro and anti fracking coalitions, this paper compares the beliefs of local government respondents to their interest group allies across two policy core beliefs: their fracking policy position and their problem perceptions. Figure 2 shows the mean policy position scores of all respondents by their organizational affiliation and with local governments separated into their coalitions Results show regardless of coalition, r espondents associated with l ocal governments have less extreme policy preferences than the respondents associated with their expected interest groups allies.

PAGE 154

141 Figure 2 Mean policy position score by organization affiliation, with local government split between pro and anti fracking coalitions

PAGE 155

142 Next, this paper compares the extremeness of problem perceptions a second policy core belief of respondent associated with local governments and the ir interest group allies. To do so, first, this paper measures the mean problem perception scores of the local government policy actors separated by their pro and anti fracking positions and the scores of their interest group allies for eac h of the 20 iss ues Then, each average score is normalized For example, if respondents associated with environmental groups indicated, on average, that the issue of Contamination of ground and surface water supplies from chemicals in hydraulic fracturing fluids perception score was 4.2, then their normalized score would be 4.2 3 = 1.2. The average normalized p roblem perception scores of respondents associated with local governments were compared to their interest group allies Second the statistical significance of the difference in mean problem perception between local governmental respondents and their allie s is tested using an analysis of variance (ANOVA) with a Bonferroni pairwise post hoc Results find policy actors from environmental groups were more extreme in 17 of the 20 issues than respondents associated with local government who had anti fracking po sitions but only one of those differences contamination of ground and surface water supplies from fracking was statistically significant at a p value < 0.10. Similarly, the respondents from the oil and gas industry were more extreme in 16 of the 20 is sues than respondents associated with local government s with pro fracking positions but only two of those diff erences ineffective monitoring by the state and the influence of the oil and gas industry over state government were statistically significan t at a p value < 0.10. See the A ppendix C Table 5 for full results

PAGE 156

143 The pairwise comparison (Table 4) provides additional information when the significant differences in mean problem perception scores are compared across all group types. First, the analys is shows that policy actors from environmental groups and oil and gas industry groups are the most different: they have statistically different problem perception scores for 90% of the issues. Second, the analysis shows policy actors from local governments have greater differences when compared to their interest group opponents, than when compared to the local government respondents in the opposite coalition. The pro fracking local government policy actors differ from the environmental policy actors on 85% of the issues, and differ with the industry policy actors on 10% of issues. The anti fracking local government policy actors differ from the oil and gas industry policy actors on 70% of the issues, and differ with the environmental policy actors on 5% of t he issues. The pro fracking and anti fracking local government policy actors differ on 70% (14/20) of the issues. A closer examination of the six common problems shows many of those which all local government respondents agreed were local level issues (Tab le 5). 45 45 The Oil and Gas and the Environm ental/Citizen groups had similar problem perceptions on two issues: A patchwork of local regulations on hydraulic fracturing and Inadequate or incomplete communication by the oil and gas industry about the risks, benefits and effects of hydraulic fracturin g to the public.

PAGE 157

144 Table 4. Percent of significantly different problem perception scores between groups, at a 0.10 level of significance. Affiliation Env. groups Oil and Gas Local Gov. P ro Local Gov. A nti State Gov. Federal Gov. Academics/ consultants Environmental groups 90% 85% 5% 80% 60% 65% Oil & gas groups 90% 10% 70% 5% 50% 25% Local Gov. Pro fracking 85% 10% 70% 0% 15% 5% Local Gov. Anti fracking 5% 70% 70% 40% 10% 30% State Government 80% 5% 0% 40% 10% 5% Federal Gov ernment 60% 50% 15% 10% 10% 0% Academics & consultants 65% 25% 5% 30% 5% 0% Table 5. Similar problem perceptions between Pro and Anti Fracking Local Government respondents. 1. Scare tactics and demonizing of the oil and gas industry by opponents of fracking. 2. A patchwork of local regulations on hydraulic fracturing. 3. Boom and bust economic cycles from natural gas development. 4. Public distrust of the oil and gas industry. 5. Conflict between mineral rights and property rights owners. 6. Burdens on local gov. services from temporary employees for well site operations.

PAGE 158

145 Overall, t he comparison of means and their extremeness provides mixed results for Hypothesis 2. While the policy actors from interest groups had an average problem perception score that was more extreme than the policy actors from local governments, there was rarely a statistical difference between the two scores. However, when the difference in scores is viewed across all groups, there are more differences between the policy actors from interest groups than between the policy actors from local governments, indicatin g the pro fracking and anti fracking local government groups may lay somewhere in between the environmental and oil and gas industry groups, on the whole. Problem perception scores: deductive and inductive clustering Next, the paper uses two cluster analyses to identify patterns in the 20 problem perception scores of local government respondents and their interest group allies that pairwise comparisons may not have found. First this paper uses a deductive approach by assuming respondents will fall i nto a pro or anti fracking coalition and setting the K means Cluster analysis to have a 2 cluster solution. Second, this paper uses an indicative clustering approach by allowing the K means analysis to determine an optimal number of clusters. T he 2 clust er K means analysis show s not all respondents associated with local governments with anti fracking and pro fracking positions fall into the expected cluster s As seen in Table 6 below, t hree of the twelve anti fracking local government respondents clustere d under while two of the 24 pro fracking local government cluster Only one industry group respondent landed in the unexpected cluster and all environmental group respondents clustered as expected. 46 46 Hierarchical 2 cluster solution using the same 20 problem perception scores show similar results

PAGE 159

146 The K means optimal solution found 3 clusters. Shown in Table 7 below, half or more of respondents associated with pro fracking local governments and anti fracking local ther non interest group respondents fell into the neutral cluster. The 3 cluster K means analysis also shows over 90% of the environmental and oil & gas groups are in their expected anti and pro fracking clusters. The cluster analyses provide a clearer pi cture of how the different groups perceive issues associated with fracking related oil and gas development and evidence for Hypothesis 2. Indeed, local government groups have less extreme problem perceptions than their interest group allies.

PAGE 160

147 Table 6. Two Cluster solution of problem perception by organization affiliation. K means Cluster Analysis (Count of cases in each) Affiliation Cluster1 (Anti) Cluster2 (Pro) Total Local Gov. Anti fracking 9 3 12 Local Gov. Pro fracking 2 22 24 Environment groups 28 0 28 Oil & g as groups 1 37 38 State Gov ernment 1 7 8 Federal Government 6 5 11 Academic & consultants 3 8 11 Total 50 82 132 Table 7. Three Cluster solution of problem perception by organization affiliation. K means Cluster (Count of cases in each) Affiliation Cluster 1 (Anti) Cluster 2 (neutral) Cluster 3 (Pro) Total Cases Local Gov. Anti fracking 6 6 0 12 Local Gov. Pro fracking 0 10 14 24 Environment groups 27 1 0 28 Oil & g as groups 0 3 35 38 Federal Government 1 7 3 11 State Government 0 3 5 8 Academic & consultants 2 6 3 11

PAGE 161

148 Conclusions and Limitations analysis of policy actor beliefs subsystem provides insight into how the beliefs of local governmental policy actors compare to the beliefs of policy actors associated with interest groups. Overall, the analysis shows policy actors affiliated with interest group s and those affiliated with government s have meaningful differences in two pol icy core beliefs their normative policy core policy preference related to a position on fracking and their policy core beliefs related to the nature of the problems of fracking In addition, the results of the comparison of policy position means, the cor respondence analysis of policy position s by organizational affiliation and the K means cluster analysis of problem definitions show respondents affiliated with local governments have less extreme policy core beliefs than respondents affiliated with intere st groups. policy actors in administrative roles will advocate for more moderate policy positions than policy actors associated with interest groups. Specifically, the result s confirm Hypothesis 1 that local governments have more moderate policy core beliefs than interest groups. Indeed, while the interest group policy actors were as sociated with the two extreme positions: to expand fracking extensively or to stop fracking. The K means 3 cluster solution of the 20 problem perception scores shows nearly half of all local government respondents regardless of their pro or anti frack ing policy position load o nto a cluster that their expected ally interest group counterparts do not. These results support Hypothesis 2 that local governments have more moderate beliefs than their interest group allies.

PAGE 162

149 The results of this paper also sup port previous arguments made within AFC studies that organizational affiliations may coopt beliefs (Sabatier, 1988) and organizational affiliation may contribute to variation of beliefs among coalition members ( Jenkins Smith & Claire, 1993). More pointedly, these results align with the idea that policy actors from governmental organizations are influenced by their interest for continued public support and that this interest superseded their policy core beliefs (Nohrstedt, 2005; Nohrstedt, 2010) 47 This paper does now, however, test the influence of organizational affiliation on beliefs, only shows an association. While the argument is made that the governmental and interest group policy actors in this survey represent their cons tituents and so may reflect those beliefs, another argument could be made that the individuals self selected into roles where they could practice their moderate or extreme beliefs. This paper has other inherent limitations due to the targeted population, data collection method used, and analyses selected. With respect to the targeted population, policy actors are a difficult population to study because no list exists from which to create a sampling frame. While snowball sampling is appropriate for identify ing populations such as this (Singleton & Straits, 2010), the sampling method may not have been exhaustive and some actors may have been excluded. To reduce the risk of underrepresenting key policy 47 interest. The choice to focus on be liefs or self interest was a difficult one as Sabatier developed the ACF. He wrestled with self interest and beliefs as two potential constructs to explain individual policy preferences and goals. Eventually he chose to focus on policy actor beliefs over t heir self interest because he felt beliefs were interest (Sabatier, 1988, p. 142). Furthermore, Sabatier recognized organization affiliation as a potential influence, via self interest, but also identified the endo geneity issue between organizational affiliation, self interest and beliefs (i.e., were views of an individual due to organizational interests, or did they choose an organization because their beliefs aligned with the mission of an organization). Despite scholars have focused on self interest as an additional individual level influence. Studies of policy actors in the Swedish nuclear subsystem found the policy preferences of policy actors from governmental organizations are led by their self interest for continued public support and that their self interest superseded policy core beliefs when formulating policy preferences (Nohrstedt, 2005; Nohrstedt, 2010).

PAGE 163

150 actors or groups of policy actors, public testimony docume nts of recent hydraulic fracturing policy debates were used to identify and then interview key policy actors from industry, government, and the industry. Next, data were collected through online surveys which may have reduced the potential for responses. Some individuals identified did not have a valid email address publicly available. For the most salient policy actors, efforts were made to contact them through the postal service. For those policy actors with a valid email address, the request to particip junk mail. In other cases, the targeted policy actor may have had technical difficulties in completing the survey. With respect to the analysis, advocacy coalitions were identified th rough a single policy core policy preference, their preference for fracking in general, rather than a series of policy core beliefs (Pierce, 2013) or more traditional methods of using policy core beliefs and a measure of coordination (Sabatier & Jenkins Sm ith, 1993). By only using the policy core policy preference to ascertain membership, pro fracking and anti fracking coalitions may have been confounded. This, in turn, could have inflated the cross over of local governmental policy actors identified in the cluster analyses of problem perceptions. For defined as policy core beliefs by the ACF (Sabatier & Weible, 2007) found the policy actors grouped differently than when their frac king preference was used. In comparison Pierce (2013), who used the same dataset, identified two coalitions using the policy core policy preference and a When onl y the 12 problem perception questions were used as part of the coalition identifica tion, there was no cross over of policy actors based on their preference to either

PAGE 164

151 stop/limit or continue/expand fracking. In other words, all policy actors who wished frack ing to be stopped/limited fell into the anti fracking coalition, while all policy actors who wished fracking to continue/expand fell into the pro fracking coalition. This leads to a question: which policy core beliefs does one choose to appropriately ident ify coalitions? Following the ACF, part of the answer is the substantive topic that the researcher uses to identify the topic. As the pairwise analysis found, local governmental policy actors with different policy core policy preferences for fracking share problem perceptions towards the more localized issues. This finding provides insight into why, under certain circumstances, policy actors who are normally opposed in a policy debate, find themselves aligning in other scenarios. Had a different policy core policy preference question been posed, the local government respondents may have landed in different coalitions. For Should local governments have more control over fracking? oups are in opposition than when the Should fracking continue different deep core belief is at stake, then perhaps the subsystem in question is also different. Substantively, this analys is contributes to our understanding of the Colorado oil and gas policy subsystem. Scholars have described similarities and differences between the compared policy actors in level subsystems (Weible & Heikkila, 2016). This paper provides information on how policy actors associated local governments fit into the policy debates alongside interest groups. The paper shows that some policy actors affiliated with local governments align with the oil and gas industry interests while other local governmental policy actors align with environmental

PAGE 165

152 interests. But, local governmental policy actors regardless of their fracking policy stance share commonalities as a group. The analyses show that when specific fracking related problems are asked local governmental actors may act together, rather than with the oil and gas industry or environmental groups

PAGE 166

153 CHAPTER V COMPARING THE RESOURCES, NETWORKS, AND POLITICAL ACTIVITIES OF LOCAL GOVERNMENTAL POLICY ACTORS TO THEIR INTE REST GROUP ALLIES Introduction : Local Governments as Policy Actors L ocal governments are an active group of policy actors in policy process es that go beyond their jurisdiction s Local government s (i.e., municipal or county governments ) c reate policy that affect broader policies and they engage in political advocacy in broad policy debates Research on the topic of local governmental influence in state and fed eral policy focusses on two areas. First, the research examines how local government s address broad policy topics through local policy making (e.g., Lutsey & Sperling, 2008; Feiock, Francis, & Kassekert, 2010; Shipan & Volden, 2006; Haider Markel, 2001). Second, the research examines how local policy decisions influence state or federal policy making (e.g., Shipan & Volden, 2006; Ghamkar & Pickerill, 2012; Riverstone Newell, 2013). However, less is known about how local governments behave i n the policy process es outside of their jurisdiction Policy process research shows local governmental representatives are active policy advocates alongside other advocate groups working to influence policy change (e.g., Sabatier, 1988; Jenkins Smith & Sab atier, 1994; Weible 2006; Koontz et al ., 2004; Blomquist, Schlager & Heikkila, 2004; Scholz & Stiftel, 2005 ; Heikkila et al 2014 ). But, there is little discussion in this research with respect to how local government al advocates and other advocate groups within the debate compare in their advocacy behaviors and ability to engage in the policy process

PAGE 167

154 Given that local governments and traditional advocacy groups, such as interest groups, have distinct differences in rules related to their expenditures and actions, and, outside of a policy debate, their primary goals are likely different, this research expects their behavior in the policy process would be different too. To test this expectation t his paper uses attributes identified by the Advocacy Coalition Framework (ACF) to examine coalitions of policy actors in contentious policy debates T he ACF is not only built to examine policy actor behavior within policy debates, but also to examine the capacity of competing coalitions to engage in the policy proces s with the goal of affecting policy change. Indeed, the ACF is a proven tool for conducting stakeholder analyses (Weible, 2006; Elgin & Weible, 2013 ). The ACF provides this paper with two direct measures to compare local governmental representatives to other policy advocates : resources and political activities. The ACF argues that resources provide advocacy coalitions with capac ity to engage in policy debates (Sabatier & Weible, 2007; Howlett, 2009; Elgin & Weible, 2013) 48 Political activities are the re affect policy change (Sabatier, 1988). The ACF provides this paper with a third way to compare local governmental representatives to other policy advocates: political ne tworks. The ACF highlights the importance of the relationships between policy actors. Certainly, a n advocacy coalition is a network of individual policy actors who coalesce to coordinate activities aimed at achieving 48 Nohrstedt (2 this case shifting policy venues) were the me chanisms by which an external event, the World Trade Center external event did indeed shift resources and cause new venues to open in the Swedish sub system, but found only the venue shift was related to policy change. The insignificant finding related to resources also highlights another argument: some resources are more important than others and should be weighted (Tsebelis, 1995, p. 301; Nohrstedt, 2 011, p. 480).

PAGE 168

155 a policy goal. In some cases, the relat ionships a policy actor or coalition has is considered a specific type of resource as it provides potential political connections or ability to mobilize the public (Weible, 2006; Sabatier & Weible, 2007; Weible 2007). For example, if a coalition has more g overnmental connections, it may have more sway in a specific policy decision. Indeed, t he network of a policy actor can influence their relative influence within a coalition (Peters, 1998; Olsson, 2011). This paper examines both the size of the network and the specific network connections of local governmental representatives to other advocates within a policy debate. Due to the lack of research that compares local governments to other advocacy groups, this work is exploratory. The paper builds expectations that the resources for advocacy, political activities, and networks, of policy actors associated local governments will be distinct from policy actors from other advocacy groups through insights provided by the ACF and from o rganizational and public management theories. 49 from which to identify and examine the policy actors within a policy debate. As such, this paper uses r elated oil and gas development policy subsystem i n 2013 to examine how local governmental representatives compare to other policy actors in a state wide debate. Within the fracking subsystem, policy actors were divided into two opposing coalitions based their policy beliefs about fracking. Those who wanted fracking T hose that want it continued or expanded (Pierce, 2013; Heikkila et al., 2014; Weible & Heikkila, 2016). The pol icy actors in the fracking subsystem utilized a range of resources for advocacy 49 The idea that policy actors may cluster along organizational boundaries is also explored within the ACF literature (Jenkins Smith & Claire, 1993; Nohrstedt, 2005; Nohrstedt, 2010)

PAGE 169

156 engaged in a variety of political activities, and interacted with different groups to achieve their policy goals (Heikkila et al., 2014; Weible & Heikkila, 2016). Finally, re presentatives from environmental organizations the oil and gas industry, and local governments made up the three largest groups in the fracking subsystem (Heikkila et al ., 2013). Therefore, for differences between local governments and advocacy actors in a contentious statewide policy debate. The remainder of this paper outline s the theoretical backing of the research through a review of the ACF and then develop s the expectation that policy actors associated with local government and those associated with other advocacy groups will differ with respect to their resources for advocacy political activities, and political networks. Then the paper describes ing subsystem. Next, variables are operationalized and analyses used for each policy actor attribute and output are explained, followed by a description of the results. The paper concludes with a discussion of the results and a reflection on the contributi ons to the literature and limitations of the study.

PAGE 170

157 Theoretical Arguments and Expectations The Advocacy Coalition Framework (ACF) The ACF provides this paper with a set of concepts to compare local government representatives to other advocates within a subsystem. C oncepts from the ACF relevant to this research include the policy subsystem, policy actors and their beliefs, advocacy coalitions, resources, political activities, and networ ks. In the ACF, the primary unit of observation for studying policy processes and the differences between policy actors is the policy subsystem The policy subsystem is defined by three attributes. A geographic region, a substantive topic, and the indivi duals involved in the topic (Sabatier, 1988). While the analyst could vary the geographic scope of the subsystem to change the level of granularity of analysis of the processes within, typical ACF studies examine subsystems bound by national, regional, or state boundaries. The individuals involved in the policy process within the subsystem are called policy actors Policy actors can be scientists, interest group representatives part of the news media, and decision makers from all levels of government. Pol icy actors are usually professionally affiliated with an organization, involved in the policy area and dedicating at least some time to influencing either directly or indirectly the politics of the subsystem. In the ACF, policy actors are differentiated fr om other citizens by the time they devote to an issue and the extent they specialize in the issue. Within a policy subsystem, policy actors form one or more advocacy coalitions. The ACF theorizes that policy actors are motivated to form coalitions with ot her policy actors to overcome individual cognitive and physical limitations on information processing and resource utilization (Sabatier & Weible, 2007). An advocacy coalition is a group of policy

PAGE 171

158 actors who share similar policy core beliefs, including pol icy goals related to a substantive topic To achieve their shared policy goas, policy actors in coalitions coordinate the use of their individual and pooled resources and strategically engage in political activit ies within the policy subsystem 50 Expectatio ns Resources The ACF identifies as important variables to consider when investigating role in a subsystem. Resources give coalitions capacity to plan and act on different strategies (Sabatier & Weible, 2007) and to process and share information (Howlett, 2009; Elgin & Weible, 2013). When two or more coalitions are engaged in a political debate, resources are used to influence policy outcomes (Jenkins Smith, 1988). Resources for political activity include finances, leadership, access to authority, access to scientific and technical information, and the ability to mobilizable supporter (Sabatier & an aggregation of its policy actor members, but does not set expectations for how resources may vary between coalitions or policy actors. Resource based theory, or the resource based view of the firm (Penrose, 1959; d sustaining competitive advantage in the market it operates (Barney, 2001; Barney, 2002; 50 The nature of interactions between advocacy coalitions within a subsystem ranges from cooperative to conflicting (Weible, 2008). In policy subsystems with a contentious substantive topic there are typically two or three conflicting coalitions (Weible, Sa batier, & McQueen, 2009), but can range between one and five (Weible, Sabatier, & McQueen, 2009). In contentious subsystems, a dominant coalition is the coalition that maintains political control over policy decisions over extended periods and acts to keep the status When there is a dominant coalition, the opposition, possibly mobilized in one or more the minority coalitions, seeks policy change to affect the status quo in ways that is congruent with their beliefs (Sabatier & Jenkins Smith 1993; Norhstedt, 2010). Minority coalitions often seek allies from outside the subsystem or take policy debates to alternative venues than where debates are traditionally held in the subsystem (Zafonte & Sabatier, 1998 citing Fritschler (1983), Baumgartner & Jones (1993), Browne (1990), and Worsham (1997)).

PAGE 172

159 Barney & Arikan, 2006). A major assumption of resource based theory is that organizations in competition have heterogeneous bundles of resources (Barney, 1991). If we apply this assumption to competition within a policy debate, it supports the expectation that different advocacy groups may have unique bundles of resources. Furthermore, a common variant on the definition of resources within resource based theory is the distinction between resources and capabilities to use the resource (H arrison et al., 2001; Makadok, 2001). While this distinction is not held by all organization strategy scholars (Barney et al 2014), the division between resources and capability is hel pful when discussing how different organization types may have distinct differences in the resources available for political advocacy. For example, local governments and advocacy organizations may have equal amounts of financial resources, but the groups m ay have different limitations (Mosely, 2010), or incentives (Tiebout, 1956), on how those resources are used for political activities related to a specific policy topic. 51 Therefore, the resource based view of the firm not only highlights resource variation at the organization level of analysis, but acknowledges the constraints on how resources are used for advocacy varies across organizational type. 52 The preceding argument leads to the general suggestion that organizations develop a unique set of resources that are beneficial for their context and interest. Given that non governmental organizations are, for the most part, not in competition with local governments for resources, and that local governments may have different resources than interest groups. 51 Mosely (2010) found certain resources correlated with advocacy H owever the research also discusses potential discrepancies in resources between nonprofit and other organizations. For example, funding streams, organi zational size, and legal constraints may explain why small nonprofits do not join in political advocacy activities 52 Heterogeneity of resources could be viewed at the organizational level of analysis. However, this work hat this level of analysis is too complicated as there are too many individual organizations in a policy subsystem, and it would not provide insights into the research question, which is focused on policy actors associated with local governments versus int erest groups.

PAGE 173

160 The refore, the expectation is: Resources of policy actors associated with local governments will be similar to each other, and different than the resources of policy actors associated with interest groups 53 Networks The connections between policy actors ar e an important facet of political advocacy and improve a policy actor s engag e ment in the policy process. The ACF argues individual policy actors collaborate and coalesce into coalitions, in part, to overcome boundedly rational nature (Sabatier, 1988; Sabatier & Weible, 2007). Through collaboration policy actors may pool their resources for advocacy and the refore increase their ability to engage in parallel processing and activities ( Sabatier, 1998; Sabatier & Weible, 2007) Logically, the larger the network of policy actors, the greater their ability is to engage in the policy process. This paper examines the size of the network of local governmental policy actors and other policy advocates. In addition to ne twork size, the specific groups or individuals that a policy actor collaborate s with also matters For example, t he ACF identifies the presence of governmental decision makers within and advocacy coalition as type of resource coalitions have for achieving political goals (Weible, 2006; Sabatier & Pelkey 1987). Weible (2006) notes that for this reason, dominant coalitions are more likely to have coalition members in position of 53 RTD also can be used to show that organizations will create a new environment as they start to compete for resources needed for policy change this new environment could be considered the subsystem. While RTD originally focused on profit m aximizing market based organizations, nonprofit and governmental sectors are not excluded from the environments and interdependencies described in the theory. Indeed, any organization that shares a need and competes for the same vital resource will become part of the environment, or social context, described by RTD. Therefore, when an organization seeks to influence a particular policy, they enter into a new competition, driven by the mutual interest to control the decision makers and their policy outputs. Organizations may enter the political process to either influence how resources they want are distributed (as seen through RTD) or to instill their beliefs in the policy decision (as seen through the ACF). Regardless, and using the language of RDT, organiz ations also create a need for resources to influence the policy process and so they must create a new environment, or social context, driven by the need for specific resources to carry out political activities and implement political strategies to achieve the organizational goals with respect to the policy change.

PAGE 174

161 authority than minority coalitions. a (1998) findings that there can be smaller networks of policy actors within advocacy coalitions, and these networks can affect the relative influence of the policy actor within their coalition. Therefore, this paper examines the network patterns within coalitions. Resource Dependence Theory (RDT) sets up an expectation that local governments may have different networks than other advocacy groups. RTD proposes organizations are dependent on their environment (i.e., external sources) for vital res ources used to pursue their organizational interests (Pfeffer & Salancik, 1978). Pfeffer and Salancik (2003) explain that ders and toward the environment in which the leaders operate. The environment, or network, in RDT is made of the organizations that hold and compete for vital resources or may control the distribution of resources through policy. In RTD, the organizations that make up a network are often interdependent on each other for resources and they act to increase their power over the source of the resource, thereby overcoming a specific dependence on other organizations (Pfeffer & Salancik, 2003). 54 RTD implies that it is not the type of resource, but its source that brings organizations into an interdependent relationship. Therefore, two organizations that do not have competing interests will not share the same environment, or network. Additionally, ACF and p olicy networks scholars argue policy actors will collaborate more with those who they share beliefs (Sabatier, 1988; Henry, L u bell, & McCoy, 2011; Elgin & Weible, 2013). While the ACF argues that actors coalesce into advocacy coalitions 54 The methods identified by RTD to overcome dependence are essentially strategies that change the relationship between the mutually dependent organizations in the environment. RTD separates the strategies into mergers and acquisitions, long term contracts, or by engaging in politics and the policy process to influence the regulatory structure guiding the resource allocation.

PAGE 175

162 based on shared policy pr eferences, variation in policy preferences and goals exist between members of the same advocacy coalition. This variation has been attributed to both the differences in their individual beliefs (Sabatier, 1988; Weible, 2006; Nohrstedt, 2010) and to the ind Jenkins Smith & Claire, 1993; Nohrstedt, 2005; Nohrstedt, 2010) There are endogeneity issues with the organization affiliation argument (Sabatier, 1988), however there is evidence that an individual may have their bel iefs or self preferences. For example, the policy preferences of governmental actors are influenced by their interest for continued public support, and that this interest s upersedes their policy core beliefs (Nohrstedt, 2005; 2010). 55 This paper uses the setup above to justify examining the network size and network pattern of policy actors within coalitions. No expectations are developed for how the size of the network will c ompare across policy actors associated with local governments and interest groups. This chapter uses RTD pursuit for resources generates their networks, and the argument that individuals collaborate with those who they sha re similar beliefs to develop the argument that local governments have different policy networks than advocacy groups. Indeed, it is plausible that the primary network local 55 s and policy preferences is self interest. The choice to focus on beliefs or self interest was a difficult one as Sabatier developed the ACF. He wrestled with self interest and beliefs as two potential constructs to explain individual policy preferences an d goals. Eventually he chose to focus on policy actor beliefs over their self interest because he felt beliefs were interest (Sabatier, 1988, p. 142). Furthermore, Sabatier recognized organization affiliation as a potential influence, via self interest, but also identified the endogeneity issue between organizational affiliation, self interest and beliefs (i.e., were views of an individual due to organizational interests, or did they choose an organization because t heir beliefs aligned with the mission of an scholars have focused on self interest as an additional individual level influence. Studies of policy a ctors in the Swedish nuclear subsystem found the policy preferences of policy actors from governmental organizations are led by their self interest for continued public support and that their self interest superseded policy core beliefs when formulating po licy preferences (Nohrstedt, 2005; Nohrstedt, 2010).

PAGE 176

163 governments and other advocacy groups is not the policy subsystem and so their re sources may come from independent sources. Further, given the varied goals and interests of local governments and interest groups, it is expected they have some dissimilar beliefs. Therefore, the expectation is: Regardless of coalition membership, the netw ork pattern of policy actors associated with local governments will be similar to each other and different than the network patterns of policy actors associated with interest groups. Political Activities This paper uses previous research that examin es mechanisms for change (Weible, Pierce, & Heikkila 2013), that identifies specific political activities rather than broad strategies as the way policy actors engage in the policy process If strategies are a plan or method for achieving a goal, then politi cal activities can be thought of as the discrete actions of political advocates use to implement a political strategy. Public management scholars note different organization types may have unique sets of the constraints on the activities they carry out du e to institutional arrangements (Rainey & Bozeman, 2000; Rosenbloom, 2015) and/or their organizational mission (Holyoke, Brown, & Henig, 2012). Governments, for example may not participate in political advocacy unless their constituents are calling for the ir action. Furthermore, a local government official may avoid an agenda item because they are legally bound through state laws or city charters (Frug & Barron, 2007). Conversely, local governments may engage in particular types of actions because it is a r equirement or they have legal standing to do so. Because local governments can implement policy, they may engage more frequently in political activities like town hall meetings or public hearings than other organization types. In a similar vein, other orga nizations may be influenced or bound by their mission (Holyoke, Brown, & Henig, 2012). For instance, a nonprofit that sees itself as community organizer may be more likely

PAGE 177

164 to arrange protest than a nonprofit whose mission is political advocacy and spends i ts time in law making forums. Therefore, this research expects the activities of policy actors associated with local governments to be similar to each other and different than the activities of policy actors associated with interest groups. In summar y, thi expectations regarding resources, networks, and activities, are presented in Table 1.

PAGE 178

165 Table 1 Research q uestion and expectations for each policy actor attribute. Research Question Attribute Expectations How do policy actors associated with local governments compare to policy actors associated with interest groups? Resources E 1: Resources of policy actors as sociated with local governments will be similar to each other, and different than the resources of policy actors associated with interest groups. Network Size No expectation Network Pattern E 2: Regardless of coalition membership, the network pattern of policy actors associated with local governments will be similar to each other and different than the network patterns of policy actors associated with interest groups. Political activities E 3: Activities of policy actors associated with local governments will be similar to each other and different than the activities of policy actors associated with interest groups.

PAGE 179

166 Research Setting In Colorado, the fracking related oil and gas development de bates emerged in the early 2000s following the improvement and combination of horizontal drilling and hydraulic fracturing techniques. With these developments, and the discovery of new oil and gas deposits, oil and gas operations drew closer to urban populations not previously accustomed to the activity. At the same time, questions related to the environmental and public safety of fracking began to question if the practice should continue. Those for fracking related development argued the economic and energy security benefits of development, while those against oil and gas regulatory body, the Colorado Oil and Gas Conser vation Commission (COGCC) began receiving numerous complaints and concerns related to the new activity. In response, in 2007, the General Assembly passed a law requiring the COGCC promulgate new rules to reconsider impacts to the environment and public hea lth, safety, and welfare (Neslin, 2008). Then, as the national attention turned toward the chemicals used in fracking (Fisk, 2013), disclosure debates, many of the same issues related to economic benefits and environmental and public health degradation were espoused (Heikkila et al., 2014). actors in the state level debates, with vary ing resources, conducting a variety of political activities, and coordinating with different groups (Heikkila et al., 2013; Weible & Heikkila, 2016). When grouped by organizational affiliation, the majority of policy actors associated with representatives from the oil and gas industry, environmental or citizen led groups, or

PAGE 180

167 local governments (Heikkila et al., 2013). The policy actors can also be divided along a normative position that either supports or opposes fracking related development (Pierce, 2013; W eible & Heikkila, 2016). With respect to local governments, there were some for, and some against, fracking related development (Heikkila et al ., 2013; Gallaher, 2015; Wilson, 2012; Dunnahoe, 2013 UOGR magazine). With these variations, similarities and dif ferences can be compared between local governments and other interest groups, and between local governments in opposing coalitions. Methods Population and Sampling The data for this study were collected by a team of researchers (including this author) thr ough a cross sectional on line survey of policy actors in the fracking related oil and gas development policy subsystems in Colorado in 2013. We used a n on probability sampling strategy because a pre made list, or other documentation, of policy actors did not exist from which to create a sampling frame (Singleton & Straits, 2010). We used of a policy actor as our operational definition. We then used a modified snowball sampling method to identify the sample population. We began our mod ified snowball sampling method with internet searches and newspaper reviews to identify salient state level oil and gas development policy debates related to fracking The policy actor s we identified in this search acted as our seed list Next, we reviewed on line newspaper reports and documents published by the policy actors on our list to identify other policy actors. Finally, we interviewed a subset of the p olicy actors and asked who should be included in the study. We categorized p olicy actors by their organizational affiliation (e. g. industry, environmenta l group, media, and scientists) to assess

PAGE 181

168 if our methods resulted in any bias toward one group type or another. We adjusted our s earch criteria as needed to insure our sample included representation f rom different organizational affiliations. The methods we used reduced the possibility of coverage error, or the omission of policy actors involved in fracking in Colorado (Singleton & Straits, 2010). These efforts created a list of 398 policy actors. 142 of 398 policy actors responded a 35.7% response rate. Operationalization In the next section I describe how I operationalized the key variables used to examine each expectation of this paper Organizational groups (IV for E 1, E 2, E 3) I used a survey questions that asked each respondent to select organization types they most closely represented to create organizational affiliation groups combined into the oil and gas industry group The combined into the environmental group survey and were removed from the study. Local, state, and federal government respondents were kept separate. Resources (Exploratory DV for E 1; Control variable for E2 and E 3) I measured esources using a survey question that asked respondent s to give the capacity their organization ha d to mobilize a list of different resource s for advocacy Responses were given on a 4 point Likert scale where 0 = No capacity, 1 = Limited capa city, 2 = Moderate capacity, and 3 = Substantial capacity

PAGE 182

169 Sciarelli, & Airken, 2012 (in Dagnino (ed), 2012)) by an organization. In other words, some resources are internalized by the organization (Lee, Lee & Pennings, 2001) leaving other resources external to the organization. Therefore, this paper followed Weible, Pierce, & survey included i nternal resources (e.g., effective leadership, financial resources ) and external resources (e.g., access to elected public officials or government officials, and support from the public ) 56 Once the two factors were identified, I calculated an a verage internal and external resource scores value for each resource w hich loaded under the respective factor. See the A ppendix D Tabl e 1 for the individual resource scores and factor loadings. Networks (Exploratory DV for E2 ) I used the same question to operationalize network size and network pattern. A survey questions asked respondents to indicate which of 12 organizations the y col laborate d or engage d with to achieve their goals related to fracking The groups included governmental agencies or organizations, various advocacy groups, the network I used the I determined the n etwork patter n each respondent collaborated with. Political Activities (Exploratory DV for E3, control for E2) Political activities were operationalized by a set of questions asking 56 This paper used Principal Component factor analysis with Varimax rotation, a confirmatory factor analysis (Harrington, 2009). The analysis gave a two factor load, which aligned with the internal/extern al resource distinction and a C

PAGE 183

1 70 organization has recently engaged in different political activities to achieve its oil and gas policy objectives. Politica l activities included: posting information or advocating online; communicating with the news media; forming and maintaining a coalition with allies; formal complaining to regulatory commissions; lobbying elected officials; participating in public meetings; generating and disseminating research reports; taking legal action; organizing or participating in public protests; and testifying at public hearings. The ten political activities questions were reduced into two average scores using a Principal Component factor analysis with Varimax rotation. Factor 1, or primary political activities, includes participating in public meetings forming and maintaining coalitions, lobbying elected officials, testifying at public hearings, and communicating with news media. Activities associated with Factor 2, or secondary political activities, include posting information or advocating online, generating and disseminating research and reports, formal complaining to regulatory commissions, taking legal action, and organizing o r participating in public protests. 57 Political activity scores were calculated using the average frequency score for each political activities that loaded under the two factors. The average frequency for the activities that loaded under Factor 1 were nearl y double that of activities that loaded under Factor 2 (see appendix). Therefore, Factor 1 is described as Primary Activity Level and Factor 2 is described as Secondary Activity Level. See Appendix D Table 2. Policy preference: Policy core belief and coal ition membership (Control Variable for E 2 and E 3) natural gas development that uses hydraulic fract uring. It should be... Stopped ; Limited ; 57

PAGE 184

171 C ontinued at the current rate ; Expanded moderately ; or Expanded extensively beliefs that project an image of how t & Weible, 2007, pg. 195). The policy preference response was transformed into a dichotomous variable following previous identification of two coalitions from these response: an anti fracking stance was determined for those who stated hydraulic fracturing should be stopped or limited. A pro fracking stance was determined for those who stated hydraulic fracturing should be continued at its current rate, expanded moderately, or expanded extensively (Pierce, 2013). 58 Years involved (Control Variable for E 2 and E 3) A demographic question, years involved in the subsystem, is used as a control variable for the activity and network comparison. The more years of involvement, a policy actor would have had more time to make connections and potentially more knowledge of or better access to resources. 58 A recent study utilizing this data set used problem perceptions and policy preference as a way t o identify coalition membership using a cluster analysis of Manhattan distances between individual survey respondents based on 2 scores: one score used problem perception score from 12 survey questions. As a result, two coalitions emerged: an anti fracking and pro analysis shows only two of 133 respondents were placed a coalition that did not align with their position on hydraulic fracturing.

PAGE 185

172 Analysis and Results Resources To test the expectation that resource types will vary across organization types, the internal and external resource scores are compared between p olicy actors from local governments and organization type using an ANOVA with a Fisher Hayter pairwise comparison s post hoc analysis. Table 2 below shows the mean scores of internal and external resources by group affiliation and a designation as to wheth er the difference between local governments and each other organization type is significant at a p value < 0.10. For external resources, results show no differences between local governments and the other organization types. For internal resources, local governments had fewer internal resources than oil and gas, environmental, and academic groups. No differences were found between local governments and other groups. 59 59 See the A ppendix D Table 3a and 3b for the full results of the ANOVA and Fisher Hayter pairwise compar ison for external r esources. See Appendix D Table 4a and 4b for the full results of the ANOVA and Fisher Hayter pairwise compar ison for internal resources.

PAGE 186

173 Table 2. ANOVA comparison of external and internal resources between local governments and other organization types. Affiliation External Resources Internal Resources Mean Std. dev Mean Std. dev Local Government (comparison) 1.69 0.64 1.36 0.65 Oil & g as i ndustry 1.67 0.51 1.91* 0.67 Environmental g roups 1.92 0.47 1.83* 0.66 Federal Government 1.47 0.48 1.67 0.61 State Government 1.67 0.61 1.69 0.62 Academics & consultants 1.19 0.63 2.03* 0.86 Overall resources 1.67 0.58 1.71 0.7 p<0.10, ** p<0.05, *** p<0.01

PAGE 187

174 Political Activities This research used two linear regression model s to test whether policy actors from similar organization types conduct similar activities and policy actors from different organization types conduct dissimilar activities. Given that organization type is a categorical variable, local government was exclude d from the model so that the differences in coefficients for organization type were in comparison to local governments. A separate model was run for each activity type ( e. g. primary and secondary activities). Political activity level i = Organization Type + Internal Resources + External Resources + Years Involved + Position on Fracking. Primary activities model. Figure 1 displays the marginal effects of each variable on a actors associated with oil & gas and environmental groups engage in more primary political activities than those associated with local governments. Internal and external resources are both positively associated with greater amounts of primary political activities. The number of lationship to their primary political activity levels. Secondary activities model. Figure 2 displays the marginal effects of each variable on secondary activity level, at a 90 % confidence interval. Only policy actors associated with environmental groups conduct statistically significant higher secondary political activity level than those from local governments. Internal resources had a positive and significant relationship on secondary political activity levels. Years involved and position on fracking ha d no significant relationship with second ary political activity levels. See Appendix D Table 5 through Table 7.

PAGE 188

175 Figure 1. Marginal effe cts on primary activities with 90% confidence intervals Figure 2. Marginal effects on secondary activities with 90 % confidence intervals

PAGE 189

176 Networks Analysis of which groups the local government al and other advocacy group respondents most commonly collaborated with show variation across advocacy groups. On average, t he most common group respondents indicated that they collaborated with was the state government. This result is expected given the fracking policy debates centered on most common collabor ators varied. For example, local government respondents noted they collaborated most often with other local governments, then the state, and then industry associations. Local governments collaborated least often with the federal government, scientist, and real estate groups. Environmental advocacy groups collaborated with other environmental groups most often, followed by state, local, and federal governments, respectively. Environmental respondents indicated they collaborated least often with industry grou ps and real estate groups. See Table 3 below collaboration patterns.

PAGE 190

177 Table 3 The rank of most often collaborator by advocacy group. Advocacy Group Collaborator Local Government State Government Federal Government Oil and Gas Industry Env ironment /Citizen Groups Academics & consultants Local Gov. 1 3 8 4 3 8 State Gov. 2 1 2 3 5 3 Reg. Gov. 4 3 5 7 6 8 Federal Gov. 10 8 1 5 8 6 Oil & Gas Industry 3 3 5 2 12 3 Industry Associations 4 1 4 1 10 2 Env. Groups 6 3 2 6 1 3 Citizen Groups 7 8 10 12 2 10 Scientists 12 3 5 8 4 1 Media 8 11 8 9 7 6 Agriculture 9 10 11 10 9 11 Real Estate 11 11 12 11 11 11 Note. 1 = most common collaborator, 12 = least common collaborator The most common collaborator for each advocacy group is highlighted in grey.

PAGE 191

178 Network size. A linear regression model was used examine the network size of policy actors across organization types. Organization type is the dependent variable of interest, while resources, years involved, and the policy actor s position on fracking are included as controls. Network size = Organization Type + Internal Resource Capacity + External Resource Capacity + Years Involved + Position on Fracking. Results show that policy actors associated with federal government, state government, and oil and gas groups have statistically larger networks nearly two more average connections than policy actors associated with local governments, at a p value <0.10, all else being equal (Figure 3). No diff erence in network size was observed between policy actors associated with environmental groups, academics, and local governments. Additionally, internal and external resource capacity have significant and positive effects on network size at a p value <0.10 No differences in network size were observed with respect to the number of years a policy actor was involved in fracking issues or their position on fracking. See Appendix D Table 8.

PAGE 192

179 Figure 3. Marginal effects on network size with 90% confidence i ntervals.

PAGE 193

180 Network pattern. This paper used a multiple correspondence analysis (MCA) to compare the network patterns of local governmental advocates to other policy advocate groups ( Benzecri 1973; Greenacre, 1984; 1993). 60 MCA is a method used to examine patterns among multiple categorical variables, and is considered a type of factor analysis (Abdi & Valentin, 2007 ; Beh & Lombardo, 2014 ) Beh and Lombardo (2014) argue MCA is a useful tool for analyzing survey data because it allows the researcher to examine the association from a number of questions with categorical responses (pg. 388). MCA results dimension is similar to the fact ors in a factor analysis. In this paper, t he MCA analysis was based on the 12 yes/no collaboration answers in the survey, and the The number of y ears a respondent indicate d they were i nvolved in frackin g was supplied as a passive, supplementary variable. In MCA, s upplementary variables do not affect the model, but they are shown in the model outputs (Stata, n.d) The correspondence analysis follows a previous ACF study by Zafonte and Sabatier (1998), whi ch examined collaboration patterns of different organization types (called classes in their research) in the San Francisco Bay Delta subsystem. Prior to the analysis, t he local government respondents were split into two sub groups based on their policy p reference toward fracking. Local government respondents who desired fracking to continue or expand were placed into pro fracking group; those who desired fracking to be stopped or limited were placed into an anti fracking group. Given that the environmenta l advocates align with the anti fracking group and the oil and gas industry 60 Correspondence analysis is appropriate for multivariate analyses with nominal data and is considered an cluster individual respondents by their organizational affiliation and collaboration choices.

PAGE 194

181 advocates align with the pro fracking group, the sub groups of local government policy actors allow this research to explore the expectation that regardless of coalition membership the policy network of policy actors associated with local governments will be similar to each other and different than the network of policy actors associated with interest groups. Prior to describing the details, consider a n MCA where the variables inc luded in the model had no patterns, or zero dimensions (i.e., factors) along which to explain the variation in the variables. In this hypothetical case, all points would fall onto a singular point In MCA, proximity between points signifies a relationship ( Beh & Lombardo, 2014, pg. 394 ). Therefore, in the hypothetical case where all points were located together, they would be considered identica l. would be that given the variables included, no relationship is found. However if th e points were divided along one more dimensions, it would signify an underlying pattern exists. In these cases, the proximity would indicate the relationship between points, and the location along the dimension would illustrate the underlying factor corres ponding to the points. MCA also provides the weight of each (i.e., the principal inertia) by providing the T of network patterns fou nd the variables were explained by three dimensions (Beh & Lombardo, 2014, pg. 127 128) The MCA visually displays how the variables are spread along each dimension factor. Therefore, to describe the results, each dimension is treated like a factor and the percent of explanation of each dimension is used to put the stratification across the dimensions into perspective.

PAGE 195

182 The MCA gave a principal inertia of 0.12 and found 85% of the variation was explained over 3 dimensions 61 Dimension 1 explains 69% Dimensio n 2 explains 11% Dimension 3, explains 4% Figure 4 displays clustering along Dimension 1 and 2 Figure 5 displays the clustering along Dimensions 1 and 3. Figure 4 shows p olicy actors associated with oil and gas, federal government, and environmental groups are more likely to answer to the collaboration question, meaning they collaborate with more groups. Conversely policy actors associated with academics, pro frackin g local governments and anti fracking local governments are more likely to answer to the collaboration question The yes/no questions appear to drive Dimension 1 and are accounting for most explanatory power of the model. These results in comparison to the network size regression results (Figure 3), partially align in that the local government groups environmental groups are further to left of the x axis than the state government, and further away from the local government groups along Dimension 1, which does not align with the results of the regression analysis in F igure 3, above. Along D imension 2 in Figure 4 the policy actors from oil and gas, pro fracking local government and federal government groups group together above the x axis, while policy actors from environmental and anti fracking local governments below the x axis. The number of years involved runs along the y axis in the appropriate order (e. g. 0 1 years at the lowest point on the y axis and 21+ years at the highest place on the y axis). This result 61 Principal inertia is the total amount of variation explained by the model. See Appendix D Table 10a and 10b.

PAGE 196

183 indicates that Di mension 2 is associated with the number of years a policy actor is involved in the subsystem. Figure 5 shows clustering along Dimen sion 1 and 3. Pro fracking local governments and anti fracking local governments cluster near along Dimension 3 they are both located below the x axis and to the right of the y axis T he oil and gas, environmental groups, and state government cluster nea r one another near the center of the axes Overall the analysis shows some of the network pattern of policy actors is associated with their organizational type, and transcends divisions based on their policy preference toward fracking. This supports the ne twork pattern hypothesis. Overall, the MCA analysis shows that the respondents collaboration answers (e. g. most their clustering patterns and observed along Dimension 1. This clustering pattern is most easily observed w hen Dimension 1 and 2 are viewed. Additionally, results show that along Dimension 2, the two local government groups in closer proximity to their interest group allies. For example, pro fracking local government respondents are closer to the oil and gas re spondents than the other anti fracking local government respondents. However, when the 3 dimensional space is examined along Dimension 3, the results show the two local government groups are near while the other advocate groups (oil and gas, environmental groups) are in close proximity to each other in a separate location in the 3 diemnsional space.

PAGE 197

184 Figure 4. Multiple Correspondence Analysis: Dimension 1 and 2

PAGE 198

185 Figure 5. Multiple Correspondence Analysis: Dimension 1 and 3

PAGE 199

186 Conclusions and Limitations how local governmental policy actors compare to policy actors associated with interest groups. Overall, results indicate local government policy actors play an active and somewhat niche role in statewide fracking debates. They engage in fewer primary political activities than industry and environmental interest groups in the state, and fewer secondary activities than environmental interest groups in the state. A dditionally, local government policy actors have similar levels of external resources, but fewer internal resources for political activity than interest group policy actors. Local government policy actors also have smaller networks than interest group poli cy actors. Of course, the level of activity of local governmental policy actors may have changed since the survey. For example, conflict over local authority continued to increase after 2013 (Gallaher, 2015). In 2013 and 2014, despite legal threats from Go vernor John Hickenlooper and legal action from industry and the COGCC, some local governments in Colorado continued to push for more control ( Rochat, 2013; Antonacci, 2014; Hirji, 2014; CBSDenver, 2013). Even though Governor Hickenlooper called a truce be tween the state, industry groups, and local governments by creating a task force in 2014 to examine the local control issue local control bills continue to surface (The State of Colorado, 2014; Bunch, 2016) While no comparison is made between the time pe riod in this study and Finally, the networks of local government policy actors are influenced by their organization type and their position on fracking. For example, along one dimension in the multiple correspondence analysis, policy actors associated with pro fracking local gov ernments and industry groups have similar networks, while policy actors associated with

PAGE 200

187 anti fracking local governments and environmental groups have similar networks. Then, along another dimension, the policy actors associated with pro fracking and anti f racking local governments clustered near each other, while the environmental and industry policy actors clustered separately. Table 3 below summarizes the results with respect to how policy actors associated with local governments compare to policy actors from interest groups.

PAGE 201

188 Table 3 Outcomes related to the resources and network expectations Attribute Expectation Outcome Resources E 1: Resources of policy actors associated with local governments will be similar to each other, and different than the re sources of policy actors associated with interest groups. Partial Support External resources : No differences observed between policy actors associated with interest groups and local governments. Internal resources: Policy actors associated with interest gr oups have more internal resources than those associated with local governments. Network Size No expectation Policy actors associated with oil and gas groups have larger networks than those from local governments. Policy Network E2 : Regardless of coalition membership, the policy network of policy actors associated with local governments will be similar to each other and different than the network of policy actors associated with interest groups. Partially Support Over 85% of the variation in a poli network is explained along three dimensions. Dimension 1 : Policy actors associated with interest groups and the federal government are other groups than policy actors associated with local governme nt, state government, and academics group. Dimension 2: Policy actors associated with pro fracking local governments have different collaboration patterns than policy actors associated with anti fracking local governments, and are in the direction of their interest group allies. Policy actors associated with environmental groups have substantially distinct collaboration patterns from other groups. Differences along Dimension 2 are associated with years involved in the subsystem. Dimension 3 : Policy actors a ssociated with local governments (both pro fracking and anti fracking ) cluster, and policy actors from interest groups (both environmental and industry) cluster. Political activities E3: Activities of policy actors associated with local governments will be similar to each other and different than the activities of policy actors associated with interest groups. Partial Support Primary activities : Policy actors associated with interest groups engage in more primary activities than those associated with local governments. Secondary activities : Policy actors associated with environmental groups engage in more secondary activities than those associa ted with local governments.

PAGE 202

189 In addition to the potential changes in local government activity since the time of this research, t his paper has inherent limitations due to the targeted population, data collection method used, and similarities between the m easures of resources and networks. With respect to the targeted population, policy actors are a difficult population to study because no list exists from which to create a sampling frame. While snowball sampling is appropriate for identifying populations s uch as this (Singleton & Straits, 2010), the sampling method may not have been exhaustive and some actors may have been excluded. To reduce the risk of underrepresenting key policy actors or groups of policy actors, public testimony documents of recent hyd raulic fracturing policy debates were used to identify and then interview key policy actors from industry, government, and the industry. Next, data were collected through online surveys which may have reduced the potential for responses. Some individuals identified did not have a valid email address publicly available. For the most salient policy actors, efforts were made to contact them through the postal service. For those policy actors with a valid email address, the request to participate in the study could have been filtered to technical difficulties in completing the survey. With respect to measurement issues, careful consideration should be taken when review ing the results of the network size model and the external resource capacity variable. Three of the six external resource questions related to access to specific groups (media, elective political officials, and government officials) and the network collabo ration question asked if policy actors collaborated with government officials (including elected officials) from federal, state, regional, or local governments, and from the media. The similarities could lead to an endogenous relationship within the OLS mo del on network size. The issue is

PAGE 203

190 partially mitigated by the composite resource score through the factor analysis, which reduces the effect of the specific questions. Further, this research did not attempt to draw conclusions of network size based external resource capacity. Th ese issues do carry over to the correspondence analysis of collaboration patterns, which are used to test hypothesis three, that regardless of coalition, similarities in networks will exist across all local governmental policy actors. Overall, these results add to the ACF by showing the attributes used to distinguish between coalitions can be used to compare policy actors associated with groups within coalitions. The research also shows the limits of the ACF and the utility of using a pplicable theories within the framework to build the more nuanced expectations between policy actors within subsystem and coalition. Further, the results build on the argument that organization type is a meaningful differentiator between policy actors ( Jen kins Smith & Claire, 1993; Nohrstedt, 2005; Nohrstedt, 2010), and that those differences transcend coalition memberships. Indeed, this research does not bring a new idea that variation among coalition members exists, but it does highlight that policy actor s may play unique roles within the subsystem. Future research should continue to identify policy actor attributes (organizational affiliation, network patterns, resources, activities) and link those to attributes that signify who is more central to both su bsystem activity and political influence ( Gais & Walker, 1991). Substantively, this research contributes to our understanding of the Colorado fracking policy subsystem. Scholars have described similarities and differences between the policy actors within the Colorado subsystem (Heikkila et al., 2013; Pierce, 2013), to policy actors in other state level subsystems (Weible & Heikkila, 2016), and the drivers of the fracking policy conflict in Colorado (Heikkila & Weible, 2017). This paper provides more

PAGE 204

191 infor mation on the role of local governments within these debates and how they compare to interest groups with respect to resources, activities, and networks. Indeed, the most in depth local level research is focused on the legal authority of local governments (Minor, 2014). Here we see local governmental actors not in the role of a policy maker, but also in the role of a policy advocate.

PAGE 205

192 CHAPTER VI CONCLUSIONS This dissertation examined policy actor beliefs and behaviors in contentious policy debates. Table 1 research question s hypotheses, and results. Overall, the results were mixed. In some areas, this research confirmed its hypotheses while in others, the identified patterns and tested hypotheses yielded unexpected results However, both the types of results provide insight into policy process theories on policy actor beliefs and behavior and give direction for future research. Summary of work C hapter 2 examined how beliefs that are chy affect beliefs lower on their belief hierarchy in Colorado state wide policy sub s ystem s Chapter 2 asked Ho w into their secondary beliefs? how involved should government be in daily life and a policy core belief the degree to which fracking should continue were evaluated for their effect on secondary belief whic h level of government should regulate four specific issues related to fracking. Two hypotheses were posed. First, policy actors who believe government should be involved less in daily life would prefer lower levels of government to regulate specific fracki ng related issues. Second, policy actors whose beliefs do not align with the policy status quo (i.e. those who wish fracking to be stopped or limited) would choose the levels of government who are n ot currently regulating the fracking related issue as thei r preferred regulator.

PAGE 206

193 Results models provided mixed support for the hypotheses. For example, the models related to broader issue s of air and water quality, showed the expected relationship posed in Hypothesis 1 that those who believe government should be more involved in daily life also desired higher levels of government to regulate the issues and those who believe government should not be involved in daily life preferred local regulators. However, for localiz ed issues of nuisance issues from the well site operations and toward government in daily life had no impact on which level of government they preferred as regu lator. The results also gave mixed support for Hypothesis 2 T he results of the two localized issue models show ed no difference in preferred regulators between those in Colorado and Texas even though the se two states regulate localized issues at different levels of government Further, even though the results of the two broader issue models were in alignment with the hypothesized expectation, the general trends seen across all four models indicate the curre nt regulatory body currently for a specific issue is not the primary factor to consider. Rather, results indicate re gulators, and the nature of the issue Chapter 3 examined the strategic behavior of venue shopping of policy actors in New state wide subsystem. Within this chapter two questions were asked. First, What factors influence the total number o f venues a policy actor shops? Three hypotheses from the venue shopping literature were posed related to this question First, p olicy actors with more resources will shop more than those with fewer resources Second, p olicy actors associated with interest groups will shop more than non interest group policy actors Third,

PAGE 207

194 p olicy actors who desire policy change will shop more than those who do not desire policy change. A single ordered logit regression model was developed to simultaneously test the three hyp otheses. sis The second research question in Chapter 3 was What factors explain specific venue shopping choices of a policy actor? Two hypotheses from the venue shopping literature were tested within this research question. First, policy actors are more likely to shop venues they perceive to be more influential in the substantive issue. Second, Policy actors are more likely to shop ve nues with decision makers the policy actors agree with and policy actors are less likely to shop venues with decision makers the policy actors disagree with. An ordered logistic regression model of venue shopping was developed and tested across six venues within the New York statewide fracking policy subsystem. Results from the six models did not support the hypotheses. However, the results showed significant relationships between or national focus) and when looking across all six models, one venue informs their shopping patterns in other venues state wide fracking policy subsystem. Chapter 4 asked How do policy actors associated with local governments compare to other groups of policy actors with respect to their beliefs ? Results indicate local governmental policy actors have i) more moderate b eliefs than policy actors associated with interest groups and ii) they have more moderate beliefs than their Chapter 5 asked, How do policy actors associated with local governments compare t o other groups of policy actors with respect to their resources for advocacy political activities and networks among

PAGE 208

195 different policy actors in the subsystem ? Chapter 5 developed three expectations: policy actors associated with local governments will be similar to other policy actors associated with local governments with respect to their resources, activities and networks, but different from policy actors associated with interest groups. The results showed partial support for each expectation. Results indicated local governmental and interest group policy actors (i.e. oil and gas industry and environmental groups) had no differences with respect to their external resources, but significant differences were seen between these two policy actor groups with respect to their internal resources. Similarly, local governmental policy actors and interest groups engaged in significantly levels of primary political activities, but only environmental interest groups engaged in different levels of secondary political activities. Finally, a multiple correspondence analysis gave a 3 dimension result and differences between local governmental policy actors and interest groups were only seen along one dimension

PAGE 209

196 Table 1 Research Summary Questions and Hypotheses Results Ch. 2: How do a policy actor's deep and policy core beliefs translate into their secondary beliefs? Localized Issues Broad er Issues Hypothesis 1: Policy actors who believe governments should be involved less in daily life prefer lower levels of government to regulate oil and gas issues. No Support Support Hypothesis 2: Policy actors whose beliefs do not align with the status quo prefer regulators at levels of government that are different than where they are currently administered. Partial Support Partial Support Ch. 3: What factors influence the total number of venues a policy actor shops? Hypothesis 1: Policy actors with more resources shop more than policy actors with fewer resources. Support Hypothesis 2: Policy actors associated with interest groups shop more than policy actors associated with non interest groups. Support Hypothesis 3: Policy actors who desire policy change shop more than those who do not desire policy change. Support Ch 3: What factors explain specific venue shopping choices of a policy actor? Hypothesis 4: Policy actors are more likely to shop venues they perceive to be more influential in the substantive issue. No Support Hypothesis 5: Policy actors are more likely to shop venues with decision makers the policy actors agree with. Policy actors are less likely to shop venues with decision makers the policy actors disagree with. No Support Ch. 4 How do policy actors associated with local governments compare to other groups of policy actors with respect to their beliefs? Hypothesis 1: Local governmental policy actors have more moderate beliefs than interest groups. Support Hypothesis 2: Local governmental policy actors have more moderate beliefs than their allies. Support Ch. 5 How do policy actors associated with local governments compare to other groups of policy actors with respect to their resources, networks, and activities? Expectation 01: Resources of policy actors associated with local governments will be similar to each other, and different than the resources of policy actors associated with interest groups. Partial Support Expectation 02: Regardless of coalition membership, the network pattern of policy actors associated with local governments will be similar to each other and different than the network patterns of policy actors associated with interest groups. Partial Support Expectation 03: Activities of policy actors associated with local governments will be similar to each other and different than the activities of policy actors associated with interest groups. Partial Support

PAGE 210

197 Contributions This dissertation contributes to the literature in four areas. The first three contributions relate to policy scholars understanding of policy actor beliefs and behaviors in contentious policy debates. The fourth contribution is substantive and relates to the topic of state level fracking debates. First, this dissertation contributes to the ACF and policy process schol ars understanding of the relationship between a core, policy core beliefs and secondary beliefs. Results from Chapter 2 support the assumption that the beliefs at higher levels of the hierarchical belief system (i.e. deep core and policy core) inform beliefs at the lowest level s of the hierarchy (i.e. secondary beliefs) However, the results indicate that the magnitude of the effect of deep core and policy core beliefs on secondary belief is not only limited, but changes b ased on the nature of the policy issue in question Indeed, the most unexpected result of Chapter 2 was how the effect of an general attitude toward government ( deep core belief) and their preference for whether fracking should be stopped/limited, continued or expanded ( policy core belief ) on their preference for who should regulate an issue (secondary belief) changed between issues with broad externalities (i.e. air and water quality) and those with localized externalities (i.e. n uisance issues and setback distances). For example, for the broad issues of air and water quality the policy actors with the general attitude that government should be more involved in daily life and the policy core belief that fracking should be stopped were most likely to prefer local government to regulate the issue In addition, the results indicate that at extreme values of the deep core belief, the deep core belief of an individual overpower s their policy core beliefs and become the main driver on t he secondary belief For example,

PAGE 211

198 the likelihood to want local regulators for individuals who do not want government involved in daily life did not change even when they held different policy core beliefs related to whether fracking should be stopped/limited, continued, or expanded. Conversely, f or issues with more localized externalities, the results show that a related to fracking is the main drive r for their secondary belief of a preference for which level of government should regulates the issue For example, the likelihood to prefer a local regulator for setback issues changed depending on their preference related to fracking, but not by their attitude toward government involved in daily life In other words, the nature of the issue, particularly the scope of the issue, mediates how the deep and policy core beliefs affect a secondary belief. Put more plainly, the results suggest spe cific policy issue inform their preference for specific policy details B ut the degree to which ideology and normative policy preferences inform specific policy preferences depends on the nature of the issue in question The dynamic between the nature of the issue and the beliefs can be used to explain why, for example, issues that divide individual s along political party lines (i.e. republican or democratic ) are often broad topics such as abortion, gay marriage ( Hillygus & Shields, 2008 ) or climate change ( Hamilton, 2010 ) Following the observed relationship between beliefs and the nature of an issue, when an individual develops a specific preference for whom to vote for, and that decision is made in the context of a broad issue they will draw upon their political ideological or moral beliefs. Rather, when their preference for who to vote for is associated with a more specific issue, like should doctors be able to prescribe marijuana to treat a disease, the individual would be expected to d raw upon their policy core belief on the subject of legalizing medical marijuana.

PAGE 212

199 These results could also inform advocacy strategies. For example, if an advocate wanted to promote a specific policy solution from a group with similar general attitudes, th ey should frame the policy solution in the context of a broad issue that appeals to those attitudes. However, if the population is rather divided on deeply held beliefs, the advocate should frame the policy solution in the context of more localized policy issue. In doing so, the advocate would be more likely to gain support across moralistic or political party lines. Chapter beliefs also informs policy scholar understands of policy core beliefs. Resu lts from Chapter 4 shows that policy actors within coalitions have significant differences in their policy core beliefs. Specifically, the results indicate policy actors associated with local governments have more moderate beliefs than their interest group allies These results hypotheses that policy actors in administrative roles will advocate for more moderate policy positions than policy actors associated with interest groups. The results also provide support to previou s ACF research that finds policy actors from similar organizational affiliation s will share similar beliefs ( Jenkins Smith & Claire, 1993; Nohrstedt, 2005; Nohrstedt, 2010). Finally, the results of the K means cluster analysis with a 3 cluster solution whi ch grpi[ed respondents based 20 problem perceptions, showed respondents affiliated with local government who identified as pro or anti fracking within the same cluster. Not only does this result show local governmental policy actors are more moderate, it i mplies that policy actors who are directly opposed to each other with respect to one substantive topic (e.g., the continuation or limitation of fracking ) may be allies in another substantive topic (e.g. whether local governments should have more control over an issue )

PAGE 213

200 The second area of contribution is on policy actor behavior. T he dissertation contributes to policy scholars understanding of policy actor behavior in contentious policy debates through the resu lts in Chapter 3 on venue shopping Specifically, contribute to the venue shopping literature by demonstrating that two methodological benefits of the ACF sub s ystem lens in three ways. The ACF approach 1) reduces the amount of unexplai ned variation due to institutional features that could affect venue shopping when the policy subsystem boundary is selected to match a state boundary is 2) inclusive of venues that span vertical and horizontal levels of government and promotes including m ore venues than typically found in venue shopping research and is 3) more inclusive of policy actors than typically included in venue shopping research. Within the analytical boundaries ources for advocacy and who desire s policy change is more likely to shop more venues within a subsystem. show policy actors associated with interest groups have a higher level of engagement in the subsystem. Indeed, those associated with interest groups are more likely to shop a greater number of venues than policy actors associated with other groups, such as local, state, or federal governments, or members o f the scientific community. second set of venue shopping models demonstrate that venue shopping choices (i.e. how often they engage with a specific venue) within an icy subsystem, are not driven by the Certainly, at the level regulatory and legislative venues had nearly half a century of making oil and gas r elated policy (IOGCC, 1994) Therefore, the large amount

PAGE 214

201 office may simply acknowledge that the policy games are played at these venues, regardless of whether the policy ac tor agrees with, or believes the venue to be influential. However, the most significant predictors of whether a policy actor will shop one venue in a subsystem are their other venue shopping patterns. This finding suggests venue level characteristics and t he associations or networks of policy actors may be more influential in individual venue selection than previously thought. For example, a lobbyist at the state legislature may not be have the skills or authority to engage a court, even if the individual f elt their policy preferences were in perfect alignment with the judge or that the court could affect policy change in the subsystem. Likewise, a lawyer may not have interest or ability to venue shop the state agency. Further, the policy actor venue shoppin g patterns exhibited in the second set of models shows the differentiation of horizontal and vertical venues may be an obfuscated way to examine the shopping tendencies of policy actors. Third, this dissertation contributes to policy scholars understandin g of local governments as policy advocates in state level politics ( Ch. 4 and 5) Overall, results indicate local government policy actors play an active and somewhat niche role in statewide fracking debates. Results from Chapter 5 show local governments h ave significant differences in their resources, network size, network pattern, and political activities when compared to most advocacy groups. Indeed, when local governmental policy actors are compared to policy actors associated with other governments, t he academic community, or interest groups, differences in each of these attributes are observed. In the fracking debates in Colorado, local governmental policy actors have fewer internal resources for advocacy than oil and gas industry and environmental policy actors. But local governmental policy actors have similar external resources when compared to all other groups. Local governmental policy actors

PAGE 215

202 engage in fewer primary political activities than policy ac tors associated with the oil and gas industry and environmental groups. Yet local governmental policy actors are only less active in their secondary activities than environmental policy actors. Finally, local governmental respondents reported to have a sm aller network of collaborators than federal and state governmental respondents and respondents from the oil and gas industry. However local governmental respondents had similar network sizes as respondents from environmental groups. The analys e s of beliefs, resources, networks, and activities in Chapter 4 and Chapter 5 find variation between local governmental actors and other advocacy groups, and variation among local governmental policy actors. The r esults from Chapter 4 show the policy core belief s of the governmental representatives vary. Because of this variation in their beliefs, local governmental policy actors are divide d into two competing coalitions. However when the mean policy core beliefs of policy actors calculated by organization type, the results show the local governmental policy actors have more moderate beliefs than their interest group allies. T he r MCA analysis of policy actor networks and the results 3 cluster analysis of policy actor beli efs show local governmental policy actors can associate with competing advocacy groups (e. g. pro fracking or anti fracking groups), but the local governmental policy actors also share significant similarities as a group in both their beliefs and their networks. T hese results support the arguments that, in spite of similar policy core beliefs, policy actors from similar organizations are more similar than their coalition member counterparts, and that those with similar be liefs associate are more

PAGE 216

203 likely to associate with one another ( Jenkins Smith & Claire, 1993; Nohrstedt, 2005; Nohrstedt, 2010; Elgin & Weible, 2013; Henry et al., 201 1 ). Lastly, this dissertation provides scholars and others interested in the topic of fra cking a substantive understanding of how policy actors act within state level subsystems of Colorado, New York, and Texas. In each state, local governmental representatives engage in the policy process in ways that are like other policy advocates. Indeed, local governments have the capacity to affect policy change beyond their boarders in more ways than generating policy of their own. Local level research shows a rich variety of policy games centered on local level policy making (Gallaher, 2015; fracktracke r.org) and non trivial response by state governments and interest groups ( Minor, 2014 ) and effects on higher level policy making ( M ufson, 2014). This research shows local governments also take on the role of the policy advocate. A role that requires furthe r inquir y While t he differences found in this dissertation related to the beliefs, resources, networks, and political activities provide insight that local governments play a unique role in state level policy subsystems, it has its limits. The questions answered and the results found in this body of work cannot reflect on whether local governments are a influential group of policy actors in the statewide fracking debates.

PAGE 217

204 Future Research I see this research expanding in three directions. First, I will expand on the results of six individual venue shopping models in Chapter 3 by exploring the relationship between a A preliminary factor analysis shows similar results to the combined six venue shopping models: policy acto rs shop within three basic groups The results from the six models appear to show a s plit along courts, state level venues, and elected venues. This indicates there may be factors beyond whether the venue is at a particular level or branch of government. O ne potential next step is use the aforementioned choice. Then, create a single multi level multinomial model. This would provide a large N and allow additional explorato ry variables and hypotheses to be tested. Second, I will continue to examine how normative b elief s translat e into specific preferences. Further research into secondary beliefs, such as t he preference for who should regulate is an area to continue to inves tigate to inform how beliefs within belief hierarchy interact The fracking debates in Colorado and Texas shows how divisive the topic definition (Sabatier & above) that the preference over which level of government should an issue is a secondary belief ( Sabatier & Weible, 2007). Previous research also identifies difficulties in defining policy core and secondary beliefs (Jenkins Smith et al 2016). Therefore, further should

PAGE 218

205 regulate across issues with varying externalities may assist in improving belief definitions in the ACF. Third, I will continue to examine the l ocal g level politics This show local governmental ac tions are more diverse than the bottom up federalism literature discusses (Gamkhar & Pickerill, 2012) In addition, the policy process literature will be informed by a better understanding of policy actor groups within the coalition and beyond traditional interest groups. Scholars inquiries into local governments will certainly have a rich test bed provided the local governments active role in issues such as climate change, immigration laws, and general defiance to current federal actions.

PAGE 219

206 REFERENCES Abdi, H., & Bera, M. (2014). Chapitres de Livre : Titre du livre : Encyclopedia of Social Networks and Mining September 2014, Springer, isbn : 978 1461461692 nique. 2007. Multiple Correspondence Analysis. In Encyclopedia of Measurement and Statistics by Neil Salkind (Ed.). Thousand Oaks, CA: Sage. Adams, G. 1981. The Politics of Defense Contracting: The Iron Triangle Transaction Publications, New Brunswick NJ. Ai., C. & Norton, E. C. 2003. Interaction terms in logit and probit models. Economics Letters 80, p. 123 129. doi:10.1016 /S0165 1765(03)00032 6. Al Sultan, K.S. (1995). A Tabu search approach to the clustering problem. Pattern Recognition, 28 (9): 1443 1 451. in the Barnett shale. Southern Rural Sociology 24(1), pp. 113 129. Antonacci, K. (November 20, 2014). Boulder, Weld counties go head to head on local contro l over oil and gas. Times Call News Retrieved from http://www.timescall.com/news/ci_26982034/boulder weld counties go head head local control Ayala The Examiner Retrieved from http://www.examiner.com/article/the story of nys s fracking moratorium Ba ker, M. B. (June 15, 2015). Denton City Council repeals fracking ban. Star Telegram Retrieved from http://www.star telegram.com/news/business/barnett shale/artic le24627469.html Barnett Shale Energy Education Council (n.d.). Legislation Retrieved from http://www.bseec.org/stories/legislation Barney, J. (2001). Resource based theories of competitive advantage: A ten year retrospective on the resource based view. J ournal of Management, 27 643 650. Barney, J. (2002). Gaining and sustaining competitive advantage 2 nd Ed. Upper Saddle River, NJ: Pretence Hall. Barney, J., & Arkian, A. (2006). The resource based view. Origins and Implications. Handbook of Strategic Management. In Hitt, M., A., Freeman, R. E., & Harrison, J. S. (eds) The Blackwell Handbook of Strategic Management. Blackwell Publishing. doi. 10.1111/b.9780631218616.2006.00009.x Baumgartner, F. R., & Leech, B. L. (2001). Interest niches and policy band wagons. Journal of Politics 63:1191 213.

PAGE 220

207 Baumgartner, F. R., & Jones B. D. (1991). Agenda Dynamics and Policy Subsystems. The Journal of Politics 53 (04), 1044 1074. Baumgartner, F. R., & Jones B.D (1993). Agendas and instability in American politics. Chicago, IL: University of Chicago Press. Baumgartner, F. R., & Jones B.D (2005). Politics of Attention: How Government Prioritizes. Chicago, IL: University of Chicago Press. Baumgartner, F R. ( 2009 ) Interest Groups and Agendas. In L. Sandy Maisel an d Jeffrey M. Berry, eds., Oxford Handbook of American Political Parties and Interest Groups New York: Oxford University Press, forthcoming, pp. 519 33. Becketti, S. ( 2013 ) In the spotlight: marginsplot. The STATA News 27 (4), p.2 3. Stata Press. ISBN 13 : 978 1 59718 132 7 Beh, E J., & Lombardo R (2014). Correspondence Analysis, edited by Eric J. Beh, and Rosaria Lombardo, John Wiley & Sons, Incorporated. ProQuest Ebook Central, http://0 ebookcentral.proquest.com.skyline.ucdenver.edu/lib/ca/detail.action?docID=1780725 Benzecri J P (1973) Vols 1 and 2. Dunod, Paris Berke, P. R., & French, S. P. (1994). The Influence of Stat e Planning Mandates on Local Plan Quality. Journal of Planning Education and Research, 13 (4), 237 250. doi:10.1177/0739456X9401300401 Beyers, J. (2009) Multilevel venue shopping in Europe: A comparative analysis of interest organizations in four EU member states. [Conference Proceedings] (Unpublished). Official URL: http://www.euce.org/eusa2009/papers.php Beyers, J. & Kerremans B (2012). Domestic Embeddedness and the Dynamics of Multilevel Venue Shopping in Four EU Member States. Governance: An International Journal of Policy, Administration, and Institutions, 25 (2), pp. 263 290. doi:10.1111/j.1468 0491.2011.01551.x Blomquist, W., Schlager, E. & Heikkila T (2004). Common Waters, Diverging Stream s: Linking institutions and water management in Arizona, California, and Colorado Washington, DC: Resources for the Future. Boekelman, K. (1992). The influence of states on federal policy adoptions Policy Studies Journal, 20 (3), p. 365 375. Briffault, R. (1990). Our Localism: Part I -The Structure of Local Government Law. Columbia Law Review. 90 (1), 1 115 Stable URL: http://www.jstor.org/stable/1122837 moratorium on hydraulic fracturing and where it stands in the threat of takings. Environmental Law Reporter. 12 : 11146 11156. Browne, W. (1990). Organized interests and their issue niches. Journal of Politics, 52 : 477 509.

PAGE 221

208 Brush, P. ( August 29, 2013 ) N Y High Court Jumps Into Fight Over Local Fracking Bans Law360. Retrieved on 2/25/2017 from https://www.law360.com/articles/468799/ny high court jumps into fight over local fracking bans Buchanan, J. M, & Tullock, G. (1962). The Calculus of Consent: Logical Foundations of Constitutional Government Indianapolis, IN: Liberty Fund, Inc. Buffardi, A. L., Pekkanen, R. J., & Smith, S. R. (2015). Shopp ing or specialization? Venue targeting among nonprofits engaged in advocacy. The Policy Studies Journal 43 (2), 188 206. DOI: 10.1111/psj.12090 Butler H. N., & Macey J. R. (1996). Externalities and the Matching Principle: The Case for Reallocating Environme ntal Regulatory Authority. Yale Law & Policy Review. 14 (2), Symposium Issue: Constructing a New Federalism: Jurisdictional Competence and Competition, 23 66. Bunch, J. (April 4, 2016). Bill to give local governments more control over fraking dies in Color ado House. The Denver Post. Retrieved on 7/17/2017 from http://www.denverpost.com/2016/04/04/bill to give local governments more control over fracking dies in colorado house/ Cairney, P. (1997) 'Advocacy Coalitions and Policy Change', in G. Stoker and J. Stanyer (eds.) Contemporary Political Studies, Nottingham: PSA. Carr, J. B. (2003). Perspectives on City County Consolidation and Its Alternatives. Chapter 1 in R. C. Feiock and J. B. Carr (Eds.), City County Consolidation and Its Alternatives: Reshaping the loca l government landscape (3 24). New York, NY: M.E. Sharpe. CBSDenver Local News Retrieved from http://denver.cbslocal.com/2013/02/26/hickenlooper threatens to sue any town city that bans fracking/ CDPHE. ( 2012 ) Colorado Department of Public Health and Environment memorandum to the COGCC rega rding stakeholder group meetings and recommendations. September 4, 2012. Retrieved from http://cogcc.state.co.us/documents/reg/Rules/2012/s etback/StakeholderGroup/Recomme ndations/CDPHE.pdf Chatterjee, S., A. S. Hadi, and B. Price. 2000. Regression Analysis by Example, 3rd edition. New York: John Wiley & Sons. Choonwoo Lee, Kyungmook Lee & Pennings J.M ( 2001 ) Internal capabilities, external networks, and performance: a study on technology based ventures. Strategic Management Journal 22 (6 7), p. 615 640. DOI: 10.1002/smj.181 Clingermayer, J. C., & Feiock, R. ( 2001 ) Institutional constraints and policy choice Albany, NY: SUNY Cobb, R.W., & Elder, C.D. (1972). Participation in American politics: The Dynamics of Agenda building Boston: Allyn & Bacon, Inc.

PAGE 222

209 COGA ( 2012 ) MYTHBUSTERS: Federal Exemptions. Colorado Oil & Gas Aossociation, published 2012. Retrieved http://www.coga.org/wp content/uploads/2015/08/18 MythBusters_FederalExemptions.pdf COGA ( 2016 ) COGA Statement on Anti Oil and Gas Measures Falling Short of Signatures Needed for Ballot. August 29, 2016. Retrieved from https://www.coga.o rg/wp content/uploads/2016/08/August 29 2016 COGA Statement on Anti Oil and Gas Measures Falling Short of Signatures Needed for Ballot.pdf COGA. ( 2014 ) Economics and Taxes: Economic Benefits. Retrieved from http://www.coga.org/index.php/FastFacts/Hydraulic_Fracturing#sthash.a32Qeu4j.dpbs Cognition: Explaining the White Male Effect in Risk Perceptio n Journal of Empirical Legal Studies, 4 (3) 465 505. Com rey A L ., & Lee H B (1992) A first course in factor analysis, 2nd ed. Hillsdale, NJ: Lawrence Erlbaum Associates. Constantelos, J. (2010). Playing the field: Federalism and the politics of venue shopping in the United States and Canada. The Journal of Federalism, 1 24. doi doi:10.1093/publius/pjq010 Constantelos, J. ( 1996 ) Multi level lobbying in the European Union: A paired sectoral comparison across the French Italian border. Regional and Federal Studies 6 (3), 28 57. Constantelos, J. ( 2004 ) The Eur opeanization of interest group politics in Italy: Business associations in Rome and the regions. Journal of European Public Policy 11 (6), 1020 1040. Constantelos, J. ( 2007 ) Interest group strategies in multi level Europe. Journal of Public Affairs 7 (1): 39 53. Constantelos, J. ( 2010 ) Playing the Field: Federalism and the Politics of Venue Shopping in the United States and Canada. Publius: The Journal of Federalism 40 (3), 460 483. doi:10.1093/publius/pjq010. Dahl, R A. ( 1961 ) Who governs? Democrac y and power in an American city New Haven: Yale University Press. Daley, D. M. & Garand J. C. (2005). Horizontal diffusion, vertical diffusion, and internal pressure in state environmental policymaking, 1989 1998. American Politics Research 33 (5), p. 61 5 644. Pennsylvania lobbying for DRBC fracking rules: New York The Morning Call. Retrieved June 26, 2012, from http://articles.mcall.com/2012 06 14/news/mc allentown marcellus fracking 20120614_1_fracking gas drilling drbc Davis, G. F. & Cobb J. A. (2010) Resource dependence theory: Past and future. Stanford's organization theory renaissance, 1970 2000: 21 42. Bingley, NY: Emerald Group.

PAGE 223

210 Douglas M. (1978). Cultural Bias Royal Anthropological Institute Occasional Paper No. 35, Royal Ant hropological Institute, London (reprinted in: M. Douglas, 1982a, In the Active Voice Routledge and Kegan Paul, London, pp. 183 254). Douglas M. (1992). Risk and Blame: Essays in Cultural Theory Routledge, London. Dunn, S ( 2013 ) Tighter emissions contr ol standards next challenge for oil and gas industry in Weld. July 18, 2013 Greeley Tribune, retrieved from htt p://www.greeleytribune.com/news/business/tighter emissions control standards next challenge for oil and gas industry in weld/ Dunnahoe, T. (December 13, 2013). Colorado communities collaborate with operators. Oil and Gas Journal, 1 (4) accessed from http://www.ogj.com/articles/uogr/print/volume 1/issue 4/colorado communities collaborate with operators.html Elgin, D. J. & Weible, C. M. (2013), A Stakeholder Analysis of Colorado Climate and Energy Issues Using Policy Analytical Capacity and the Advocacy Coalition Framework. Review of Policy Research, 30 : 114 133. doi: 10.1111/ropr.12005 Elgin, D (2014). Examining the role of resources, beliefs, and behavior in the policy process: A study of Colorado climate and energy politics and policy Dissertation, University of Colorado at Denver 2014, 272; 3667223 Environmental Engineering and Consulting, Inc. (n.d.). A Brief History of Hydraulic Fracturing. Retrieved from http://www.eecworld.com/services/258 a brief history of hydraulic fracturing Erickson, D. & Ansoetter, M. (2011). DRBC prop oses regulations on hydraulic fracturing in the Delaware river basin. Association of Corporate Counsel. Retrieved June 27, 2012, from http://www.lexology.com /library/detail.aspx?g=59addae2 0f1f 44ef 9414 77cd6ef3bcc7 Fehling, D ( 2015 ) How Texas Challenges the Power of Cities and Their Citizens. StateImpact: A reporting project of NPR member stations February 12, 2015, retrieved from https://stateimpact.npr .org/texas/2015/02/12/how texas challenges the power of cities and their citizens/ Feiock, R. C. (2002). A Quasi Market framework for Local economic development competition. Journal of Urban Affairs 24: 123 42. Feiock, R. C. (2004). Metropolitan Governanc e: Conflict, Competition and Cooperation Washington D.C.: Georgetown University Press. Feiock, R., Francis, & N., Kassekert, T. ( 2010 ) Explaining the Adoption of Climate Change Policies in Local Government. Paper prepared for the Pathways to Low Carbon C ities Workshop, Hong Kong, China. Fisk, J. M. (2013). The right to know? State politics of fracking disclosure Review of Policy Research 30, pg. 345 365. Food & Water Watch. ( 2015 ) The Urgent Case for a Ban on Fracking. Washington, DC: The Food Water Watch.

PAGE 224

211 Fracktracker. 2014. Fracking Bans and Moratoria in New York. Retrieved from http://www.fractracker.org/map/us/new york/moratoria/ Francis, N. & Fe iock, R. (2011). A guide for local government executives on energy efficiency and sustainability: Conserving Energy and the Environment Series. Published by the IBM Center for The Business of Government. Retrieved from http://www.businessofgovernment.org/sites/default/files/A%20Guide%20for%20Local %20Government%20Executives%20on%20Ener gy%20Efficiency%20and%20Sustainab ility.pdf Fritschler, A. L. (1983). Smoking and Politics, 3 rd edn. Englewood Cliffs, NJ: Prentice Hall. Frug., G. E. & D. J. Barron. (2008). City Bound: How states stifle urban innovation Ithaca: Cornell University Press Furlong, S. R. (1997). Interest Group influence in rulemaking. Administration and Society, 29 325 47. Furlong, S. R., & Kerwin, C. M. (2005). Interest group participation in rule making: A decade of change. Journal of Public Administration Research and Theory, 15 (3), 353 370. Galbraith, K. (Dec. 10, 2012). Railroad Commission Preparing Rule Changes for Oil and Gas Drilling. The Texas Tribune. Retrieved from https://www.texastribune.org/2012/12/10/texas regulators prepare major drilling rule chang/ Gallaher, S. (2015 September 9, 2015 by the School of Public Affairs University of Colorado Denver. Gallaher, S., Pierce, J. J., Weible, C. M., Kagan, J., Heikkila, T., & Summary Report of Perceptions of the Politics and Regulation of Unconventional Shale by the School of Public Affairs University of Colorado Denver. Gamkhar, S., & Pickerill, J. M. ( 2012 ) The State of American Federalism 2011 2012: A Fend for Yourself and Activist Form of Bottom Up Federalism. Publius: The Journal of Federalism 42 (3), pp. 357 386. Gastil, J., Knobloch, K., Kahan, D., & Braman, D. ( forthcoming 2016 ) Participatory policymaking across cultural cognitive divides: Two tests of cultural biasing in public form design and deliberation. Public Administration. Retrieved from http://sites.psu.edu/citizensinitiativereview/wp content/uploads/sites/23162/2015/01/Participatory Policymaking .pdf Gerring, J. ( 2007 ) Case Study Research: Principles and practices. Cambridge: Cambridge University Press. Goggin, M., (1987). Policy Design and the Politics of Implementation: The Case of Child Health Care in the American States. Knoxville, TN: The University of Ten nessee Press.

PAGE 225

212 Gottlieb, C. ( 2012 ) Regulating natural gas development through local planning and land use control. New York Zoning Law and Practice Report, 12 (6), 1 10. Retrieved 2/24/2017 from https://www.scribd.com/document/139432606/New York Zoning and Fracking Gould, W. ( 2000 ) sg124. Interpreting logistic regression in all its forms. Stata Technical Bulletin 53: 19 29. Granovetter, M. S. (1973). The Strength of We ak Ties. American Journal of Sociology, 78 (6), 1360 1380. Grant, W. (1978) 'Insider groups, outsider groups and interest group strategies in Britain', University of Warwick Department of Politics Working Party no. 19. Ground Water Protection Council and ALL Consulting. (2009). Modern shale gas development in the United States: A primer Prepared for the U.S. Department of Energy, Office of Fossil Energy, and the National Energy Technology Laboratory. DOE Award No. DE FG26 04NT15455 (April). Haider Markel, D. ( 2001 ) Policy diffusion as a geographical expansion of the scope of political conflict: Same sex marriage bans in the 1990s. State Politics & Policy Quarterly 1 (1), pp. 5 26. Hair J F Tatham R L Anderson R E ., & Black W (1998) Multivariate data a nalysis (Fifth Ed.) Prentice Hall:London. Hakim, D. (June 13, 2012). Cuomo Proposal Would Restrict Gas Drilling to a Struggling Area. New York Times. Retrieved June 22, 2012, http://www.nytimes.com/2012/06/14/nyregion/hydrofracking under cuomo plan would be restricted to a few counties.html?smid=pl share Hakim, D. (September 30, 2012). Shift by Cuomo on gas dr illing prompts both anger and praise. The New York Times. Retrieved from http://www.nytimes.com/2012/10 /01/nyregion/with new delays a growing sense that gov andrew cuomo will not approve gas drilling.html?_r=0 Hall, R., & Deardorff, A. (2006). Lobbying as Legislative Subsidy. The American Political Science Review, 100 (1), 69 84. Retrieved from http://www.j stor.org/stable/27644332 Hamilton, L. C. ( 2009 ) Statistics with STATA: Updated for Version 10 Belmont: Brooks/Cole. Hamilton, L. C. (2011). Education, politics and opinions about climate change evidence for interaction effects. Climate Change 104 (2), pp. 231 241. Hassat, K. A. & Mathur, A. (April 4, 2013). Benefits of hydraulic fracking. American Enterprise Institute. Retrieved from https://www.aei.org/publication/benefits of hydraulic fracking/ Hayduk, L. (1987). Structural equation modeling with LISREL: essentials and advances The Johns Hopkins University Press: Baltimore.

PAGE 226

213 Hayduk, L. (1990). Should model modifications be oriented toward improving data fit or encouraging cre ative and analytical thinking? Multivariate Behavioral Research, 25 (2), 193 196. Hayduk, L. (1996). LISREL issues, debates, and strategies (Baltimore: The Johns Hopkins University Press). Heikkila, T, Gallaher, S., Pierce, J., Kagan, J., Weible, C., & Crow D (2013) Understanding a Period of Policy Change: The Case of Hydraulic Fracturing Disclosure Policy in Colorado Presented at the Midwest Political Science Association conference in April, 2013. Heikkila, T., & Weible, C.M. ( 2017 ) Unpacking the inten sity of policy conflict: a study of Policy Sciences DOI 10.1007/s11077 017 9285 1 Heikkila, T., Pierce, J. J., Gallaher, S., Kagan, J., Crow, D. A., & Weible, C. M. (2014a). Understanding a period of policy change: The ca se of hydraulic fracturing disclosure policy in Colorado. Review of Policy Research 31 (2), 65 87. Heikkila, T., Weible, C. M., Pierce, J.J., Gallaher, S., Kagan, J., & Blair, B. (2014b). A Summary Report of the Politics of Shale Gas Development and High V olume Hydraulic Fracturing in New York. Published April, 2014 by the School of Public Affairs University of Colorado Denver. Heikkila, T., Weible, C. M., Pierce, J.J., Gallaher, S., & Kagan, J. (2013). A Summary Report of Perceptions of the Politics and Regulation of Hydraulic Fracturing in Colorado. Published October, 2013 by the School of Public Affairs University of Colorado Denver. Henry, A. D., Lubell, M., & McCoy, M. (2011). Belief systems and social capital as drivers of policy network structure: T he case of California regional planning. Journal of Public Administration Research and Theory 21 (3), pp. 419 444. Hillygus, D. & Shields, T. (2009). The Persuadable Voter: Wedge Issues in Presidential Campaigns PRINCETON; OXFORD: Princeton University Pres s. Hirji, Z. (December 3, 2014). Boulder Ups the 'Anti' in the Fracking Game. Inside Climate News Retrieved from, http://insideclimatenews.org/news/20141203/boulder ups anti fracking game Hojnacki, M, Kimball, D. C., Baumgartner, F. R., Berry, J, M., & Leech, B. L. (2012). Studying organizational advocacy influence: Reexamining interest group research. Annual Review of Political Science, 15, 379 99. Hojnacki, M., & Kimball, D. (1998). Organized Interests and the Decision of Whom to Lobby in Congress. The American Political Science Review, 92 (4), 775 790. Retrieved from http://www.jstor.org/stable/2586303 Holyoke, T. T. (2003). Choosing battlegrounds: Interest group lobbying across multiple venues. Political Research Quarterly 56 pp. 325 36.

PAGE 227

214 Holyoke, T. T., Brown, H., & Henig, J. R. (2012). Shopping in the Political Arena: Strategic State and Local Venue Selection by Advocates. State and Local Government Review, 44 (1), 9 20. doi:10.1177/0160323X11428620 Hostetter, L. (May 21, 2014). Local control of oil and gas issues is best choice for Colroado. The Burlington Record. Guest Column Retrieved from http://www.burlington record.com/guest columns/ci_25808084/local control oil and gas issues is best Howlett, M. (2009). Political Analytical Capacity and Evidence based Policy making: Lessons from Canada. Canadian Public Administration, 52 (2), pp.152 175. Hoyle, R. H. (2012). Handbook on structural equation modeling The Guilford Press. New York, NY. Hundley Jr., N. (1986) The West against it self: The Colorado River an institutional history. In New Courses for the Colorado. In the Matter of Changes to the Rules of the Oil & Gas Conservation Commission of the State of Colorado to Consider Hydraulic Fracturing Disclosure Rules: Hearings before the Oil and Gas Conservation Commission of the State of Colorado. 5. (Testimony of Mike Watts). IOGCC: Interstate Oil and Gas Compact Commission. ( 1993 ) Texas State Review: IOGCC/EPA State Review of Oil and Gas Exploration and Production Waste Managemen t Regulatory Programs. Oklahoma City, OK: IOGCC Retrieved June 22, 2012, from www.strongerinc.org IOGCC: Interstate Oil and Gas Compact Commission. ( 1994 ) New York State Review: IOGCC/EPA State Review of Oil and Gas Exploration and Production Waste Management Regulatory Programs. Oklahoma City, OK: IOGCC Retrieved June 22, 2012, from www.strongerinc.org Jann, B. 2 013. Predictive Margins and Marginal Effects in Stata. Presentation given at the 11 th German Stata Users Group meeting, Postdam, June 7, 2013. Jenkins Smith, H. C. & P. A Sabatier. (1994). Evaluating the Advocacy Coalition Framework. Journal of Public Pol icy, 14 (2), 175 203. Jenkins Smith, H. C. (1988). Analyti cal Debates and Policy Learning : Analysis and Change in the Federal Bureaucracy. Policy Sciences, 21 (2/3), 169 211. Jenkins Smith, H., Silva, C. L., Gupta, K. & Ripberger, J. T. (2014), Belief Syst em Continuity and Change in Policy Advocacy Coalitions: Using Cultural Theory to Specify Belief Systems, Coalitions, and Sources of Change. Policy Studies Journal, 42 (4), pp. 484 508. doi:10.1111/psj.12071 Jenkins Smith, H.C. Nohrstedt, D., Weible, C. M. & Sabatier, P. A. (2014). The advocacy coalition framework: Foundations, evolution, and ongoing research. In Theories of the Policy Process by Sabatier, P. A. and Weible C. M. (eds). Boulder, CO: Westview Press.

PAGE 228

215 Jones, B. D. (2001). Politics and the archi tecture of choice. Chicago, IL: University of Chicago Press. Kahan, D.M., Braman D. Gastil J. Slovic P., & Mertz C.K. ( 2007 ) Culture and Identity Protective Karch A. (2012). Vertical Diffusion and the Policy Making Process: The Politics of Embryonic Stem Cell Research. Political Research Quarterly, 65 (1), 48 61. doi:10.1177/1065912910385252 Kelloway, E. K. (1998). Using LISREL for Structural Equation Modelling: A R Guide Thousand Oaks, CA: Sage Publications. Kiersz, A., & Walker, H. (November, 3, 2014). These charts show the political bias of workers in each profession. The Business Insider. Retrieved from http://verdantlabs.com/politics_of_professions/ ; http://www.businessinsider.com/charts show the political bias of each profession 2014 11 Kingdon, J.W. (1984). Agenda, Alternatives and Public Policies. Boston, MA: Little, Brown and Company. Kirk, Roger E. (1998) Experimental Design: Procedures for the Behavioral Sciences, Third Edition. Monterey, California: Brooks/Cole Publishing. ISBN 0 534 25092 0. Koontz, T. S., Steelman, T. A., Carmin, J., Kormacher, K. S., Moseley, C., & Thomas C. W. (2004). Collaborative Environmental Management: What roles for government? Washington, DC: Resources for the Future. Krause, R. M. (2011). Policy Inno vation, Intergovernmental Relations, and the Adoption of Climate Protection Initiatives By U.S. Cities. Journal of Urban Affairs, 33 (1), 45 60. doi:10.1111/j.1467 9906.2010.00510.x Kriesi, H. & Jegen, M. (2001). The Swiss energy policy elite: The actor con stellation of a policy domain in transition. European Journal of Political Research, 39 : p. 251 287. Kubler, D. (2001). Understanding policy change with the advocacy coalition framework: An application to Swiss drug policy. Journal of European Public Poli cy, 8 (4), 623 641. Kwon, M., Berry, F., & Feicok, R. 2009. Understanding the adoption and timing of economic development strategies in US cities using innovation and institutional analysis. Journal of Public Administration Research and Theory, 19 967 988, Layzer, J. A. (2008). The Environmental Case: Translating Values into Policy 3rd ed. Washington, DC: CQ Press. Ley, A. J. (2016). Vested Interests, Venue Shopping, and Policy Stability: The Long Road to te Valley. Review of Policy Research, 33 (5) 506 525. doi: 10.1111/ropr.12190

PAGE 229

216 Ley, A. J., & Weber, E. P. (2015). The adaptive venue shopping framework: how emergent groups choose environmental policymaking venues. Environmental Politics,24 (5), 703 722. http://dx.doi.org/10.1080/09644016.2015.1014656 Lijphart, A. (1968). The Politics of Accommodation: Pluralism and Democracy in the Netherlands University of California Press: Berkley, CA. Lubell, M., Henry, A. D., & McCoy, M. (2010). Collaborative institutions in an ecology of games. American Journal of Political Science, 54(2), 287 300. up climate change mitigation policy. Energy Policy 36 (2), pp. 673 685. Mahoney, C. & Baumgartner, F. R. ( 2009 ) Converging perspective on interest group research in Europe and America. Paper prepared for the European Union Studies Association bi annual meeting. Marina del Rey, LA, CA, USA. April 23 26, 2009. Makadok, R. (2001), Toward a Synthesis of the Resource Based View and Dynamic Capability Views of Rent Creation. Strategic Management Journal 22(5), 387 401). Maloney, W A., Jordan, G, & McLaughlin, A. M. (1994). Interest groups and public policy: The in sider/outsider model revisited. Journal of Public Policy 14:17 38. DC Bureau Natural Resources News Service Retrieved June 26, 2012, from http://www.dcbureau.org/201201116933/natural resources news service/n y gov andrew cuomo sidesteps natural gas hydrofracking controversies.html m etrowyn.com Retrieved June 26, 2012, from http://www.metrowny.com/weekly_features_columns/170 Life_Happens_Consider_the_cost_of_hydrofracking.html May, P. ( 1996 ) Policy design and discretion: state oversight of local building regulation paper presented at APSA 1996. Mazey, S., & Richardsons, J (2001). Interest groups and EU policy making. In European Union: Power and policy making ed. Jeremy Richardson. London: Routledge. McBeth, M. K., Shanahan, E. A ., Arnell, R. J., & Hathaway, P. L. (2007). The Intersection of Narrative Policy Analysis and Policy Change Theory. Policy Studies Journal, 35 (1), 87 108. doi:10.1111/j.1541 0072.2007.00208.x McKay, A.M. (2011). The decision to lobby bureaucrats. Public Choice 147. Pp.123 38. McQuide, B. S. (2010). Interest Groups, Political Institutions and Strat egic Choices: What Influences Institutional Lobbying Strategies? Prepared for presentation at the 2010 American Political Science Association Conference, Washington, D.C., September 2 4, 2010.

PAGE 230

217 Meijerink, S. (2008). Explaining Continuity and Change in Inter national Policies; Issue linkage, venue change, and learning on policies for the river Scheldt estuary 1967 2005. Environment and Planning, 40 (4), 848 866. Meyer, A. (2012). Get the Frack Out of Town: Preemption Challenges to Local Fracking Bans in New Yor k. Columbia Journal of Law: Field Reports. Retrieved June 26, 2012, from http://www.columbiaenvironmentallaw .org/articles/get the frack out of town preemption challenges to local fracking bans in new york Mills, C. W. (1956). The Power Elite Oxford: Oxford University Press. Minor, J. (2014). Local Government Fracking Regulations: A Colorado Case Study. The St anford Law Journal, 33 (1), p. 61 122. Moe, T. M. (2015). Vested interests and political institutions. Political Science Quarterly 130, 277 318. Mosely, J. E. ( 2010 ) Organizational Resources and Environmental Incentives: Understanding the Policy Advocacy Involvement of Human Service Nonprofits. Social Science Review, 84 (1) p 57 76. Mufson, S. ( December 18, 2014 ) fracking ban. The Washington Post. Retrieved on 2/25/2017 from https://www.washingtonpost.com/news/wonk/wp/2014/12/18/heres the grassroots political story behind the new york fra cking ban/?utm_term=.9c3895635de4 Nagel, P. (2006). Policy Games and Venue Shopping: Working the stakeholder interface to broker policy change in rehab services. Australian Journal of Policy Analysis, 65 (4), p. 3 16. Neslin. D. (March 31, 2008). Letter by David Neslin to the Colorado Oil and Gas Conservation Commissioners regarding the draft rules. Published on the COGCC website Neslin, D. (May 11, 2009). Understanding the COGCC Rulemaking Presentation at the IOGCC Mid Year Summit. May 11 13, 20 09. Anchorage, Alaska. Retrieved from http://iogcc.ok.gov/Websites/iogcc/Images/2009%20Midyear%20Presentations/Neslin% 202009%20I OGCC 051309 FINAL.pdf Nohrstedt, D. (2005). External shocks and Policy change: Three Mile Island and Swedish nuclear energy policy. Journal of European Public Policy 12 :1041 59. Nohrstedt, D. (2010). Do advocacy coalitions matter? Crisis and change in Sw edish Nuclear Energy Policy. Journal of Public Administration Research and Theory, 20 (2): 309 333. Nohrstedt, D. (2011). Shifting resources and venues producing policy change in contested subsystems: A case study of Swedish Signals Intelligence Policy. Policy Studies Journal 39 (3) 461 484. North, D. ( 1984 ) Transaction costs, Institutions, and Economic History. Journal of Institutional and Theoretical Economics Bd. 140, H. 1., pp 7 17.

PAGE 231

218 North, D. (1990). Institutions, Institutional Change and Economic Performance Cambridge: Cambridge University Press. Northrup, C. (July, 5, 2014). The (Real) History of New York State Frack Bans. No Fracking Way: Speaking out to protect our communities. Retrieved from http://www.nofrackingway.us/2014/07/05/the real history of new york state bans/ NYDEC (New York Department of Environmental Conservation). (2009). New York State Oil, Gas, and Mining Resources, 2009 Annual R eport. NYDEC (New York Department of Environmental Conservation). (2010). New York State Oil, Gas, and Mining Resources, 2010 Annual Report. NYDEC (New York Department of Environmental Conservation). (2011). Revised Draft Supplemental Generic Environment al Impact Statement on the Oil, Gas and Solution Mining Regulatory Program: Well Permit Issuance for Horizontal Drilling and High Volume Hydraulic Fracturing to Develop the Marcellus Shale and Other Low Permeability Gas Reservoirs. The New York Department of Environmental Conservation. NYDEC (New York Department of Environmental Conservation). (2012). Division of Mineral Resources Environmental Assessment Form (EAF) for Well Permitting. Retrieved June 26, 2012, from http://www.dec.ny.gov/energy/1777.html NYDEC (New York Department of Environmental Conservation). (2012). Division of Mineral Resources Statewide Staffing Retrieved June 26, 2012, from http://www.dec.ny.gov/about/34315.html#Mined NYDEC (New York Department of Environmental Conservation). (2012). Generic Environmental Impact Statement on the Oil, Gas and Solution Mining Regulatory Program. Retrieved June 26, 2012, http://www.dec.ny.gov/energy/45912.html NYDEC (New York Department of Environmental Conservation). (2012). Landowner's Guide to Oil & Gas Leasing. Retrieved June 26, 2012, from http://www.dec.ny.gov/energy/1553.html NYDEC (New York Department of Environmental Conservation). (2012). Marcellus Shale Retrieved June 26, 2012, from http://www.dec .ny.gov/energy/46288.html NYDEC (New York Department of Environmental Conservation). (2012a). New York's Role in the Delaware River Basin Commission (DRBC) (2012). Retrieved June 26, 2012, from http: //www.dec.ny.gov/lands/48454.html Olsson, Jan. (2011). The Power of the Inside Activist: Understanding Policy Change by Empowering the Advocacy coalition Framework (ACF). Planning Theory & Practice 10 (2), pp. 167 187. Osborne. D. (1988). Laboratories of democracy Boston: Harvard Business School Press.

PAGE 232

219 Ostrom, E. (1990). Governing the Commons: The Evolution of Institutions for Collective Action Cambridge: Cambridge University Press. Ostrom, E. (2005). Understanding Institutional Diversity. Princeton, NJ: Princeton University Press. Ostrom, V., Tiebout, C. M., & Warren, R. (1961). The organization of government in metropolitan areas: A theoretical inquiry. The American Political Science Review 55:831 42. Papillon, M. 2011. Adapting Federalism: Indigenous Multilevel Governance in Canada and the United States. Publius, 42 (2), 289 312. doi: 10.1093/publius/pjr032 Peffley, M. A., & Hurwitz, J. (1985). A hierarchical model of attitude constraint. American Journal of Political Science, 29 (4), 871 890. Penrose, E. 1959. The theory of the growth of the firm. Oxford. Oxford University Press. Peters, G. (1998). Policy networks: myth, metaphor and reality, in: D. Marsh (Ed.) Comparing Policy Networks Peterson, P. (1981). City Limits Chicago: University of Chicago Press Peterson, R. (2000). A Meta Analysis of Variance Accounted for and. Factor Loadings in Exploratory Factor Analysis. Marketing Letters v11(3), p. 261 275. Pfeffer, J. & Salancik G. R. (1978). The External Control of Organizations: A Resource Dependence Perspective New York, NY, Harper and Row. Pfeffer, J., & Salancik, G.R. (2003). The External Control of Organizations: A Resource Dependence Perspective. Stanford, CA. Stanford University Press. Pierce, J. (2013). Po litical advocacy and natural resource development. Submitted to Society and Natural Resources Pierce, J. J., Kagan, J., Heikkila, T., Weible, C. M., & Report of Perceptions of the Politics and Regulation of Hydraulic Fract uring in Colorado Denver. Pierson, P. (2000). Increasing Returns, Path Dependence, and the Study of Politics. The American Political Science Review, 94 (2), 251 267. doi: 10.2307/2586011 Pierson, P. (2004). Politics in Time: History, Institutions, and Social Analysis. Princeton University Press Pralle, S. (2003). Venue Shopping, Political Strategy, and Policy Change: The Internationalization of Canadian Forest Advocacy. J ournal of Public Policy, 23 (3), 233 260. doi:10.1017/S0143814X03003118 Pralle, S. (2006). Branching Out, Digging In: Environmental Advocacy and Agenda Setting. Georgetown University Press.

PAGE 233

220 Agenda Setting in C anadian, 34 (2), 171 194. Princen, S. & B. Kerremans. ( 2008 ) Opportunity Structures in the EU Multi Level System. In Interest Group Politics in Europe: Lessons from EU Studies and Comparative Politics by Beyers, J. Eising, R., and W. A. Maloney (eds). New York: NY: Routledge. ProPublica. (2009). How Big is the Gas Drilling Regulatory Staff in Your State? Pro Publica Journalis in Public Interest. Retrieved June 22, 2012, from http://projects.propublica.org/gas drilling regulatory staffing/ Pross, P. A. ( 1993 ) First world interest groups ed. Clive S. Thomas, 67 79. Westport, CT: Greenwood Press. Railroad Commission of Texas. (December 13, 2011). Railroad Commissioners Adopt One of Nation's Most Comprehensive Hydraulic Fracturing Chemical Disclosure Requirements. Retrieved from http://www.rrc.state .tx.us/pressreleases/2011/121311.php Railroad Commission of Texas. (N.d). Oil and Gas FAQs. Retrieved from http://www.rrc.state.tx.us/about us/resource center/faqs/oil g as faqs/ Railroad Commission of Texas. (November, 20 2013). Barnett Shale Information. Retrieved from http://www.rrc.state.tx.us/barnettshale/index.php Railroad Commission of Texas. (September 30, 2011). Public Hearing on Draft Hydraulic Fracturing Disclosure Rule. Retrieved from http://www.rrc.state.tx.us/pressreleases/2011/093011.php Rainey, H., & Bozeman, B. ( 2000 ) Comparing public and private organization: Empirical Research and the Power of A Priori. Journal of Public Administration Research and Theory. 10 (2): 447 470. Richardson, N., Gottlieb, M., Krupnick, A., & H. Wiseman. (2013). The State of State Sha le Gas Regulation Washington, DC: Resources For the Future. Richardson, V. (March 25, 2014). Ballot proposal would stop localities from exceeding state authority on fracking. The Colorado Observer. Retrieved from http://thecoloradoobserver.com/2014/03/ballot proposal would stop localities from exceeding state authority on fracking 3/ Riley, T. ( 2007 ) Wrangling with u rban wildcatters: Defending Texas municipal oil and gas development ordinances against regulatory takings challenges. Vermont Law Review, 32 : 351 407. Ripberger J. T., Gupta, K., Silva, C. L. and Jenkins Smith, H. C. (2014), Cultural Theory and the Measurement of Deep Core Beliefs Within the Advocacy Coalition Framework. Policy Studies Journal, (42) 4, pp. 509 527. doi:10.1111/psj.12074 Riverstone Newell, L. ( 20 12 ) Bottom Up Activism: A Local Political Strategy for Higher Policy Change. Publius: The Journal of Federalism 42 (3), pp. 401 402.

PAGE 234

221 Riverstone Newell, L. ( 2013 ) The Diffusion of Local Bill of Rights Resolutions to the States. State and Local Government Review 45 (1), pp. 14 24. Rochat, S. (July 11, 2013). COGA brings state into fracking lawsuit against Longmont. Times Call retrieved from http:// www.timescall.com/ci_23642683/coga brings state into fracking lawsuit against longmont Rochefort, D. A., & Cobb, R. W. (1993). Problem definition, agenda access, and policy choice. The Policy Studies Journal, 21 (1), pp. 56 71. Rodrguez, G. ( 2017 ) Gener alized Linear Models: A Note on Interpreting Multinomial Logit Coefficients Princeton University. Website http://data.princeton.edu/wws509/stata/mlogit.html Accessed 3/10/2017. Rosenbloo m, D. H. (2015). The Public Context. In M. E. Guy & M. M. Rubin (Eds.), Public administration evolving: From foundations to the future (1 18). New York, NY: Routledge. Rousseau, J. (2012). The major political writings of Jean Jacques Rousseau. Translated a nd edited by John T. Scott. Chicago, IL: The University of Chicago Press. Sabatier, P. A. (1988). An advocacy coalition framework of policy change and the role of policy oriented learning therein. Policy Sciences, 21, 129 168. Sabatier, P. A. (1998): The advocacy coalition framework: revisions and relevance for Europe, Journal of European Public Policy 5:1, 98 130 Sabatier, P. A., & Pelkey, N. (1987). Incorporating multiple actors and guidance instruments into models of regulatory policymaking: An advocac y coalition framework. Administration and Society 19 (2), pp. 236 63. Sabatier, P. A., & Weible, C. M. (2007). The Advocacy Coalition Framework: Innovations and Clarifications. In P. A. Sabatier (Ed.), Theories of the Policy Process (2nd ed., pp. 189 2207). Boulder: Westview Press. Sabatier, P. & Jenkins Smith, H. (1993). Policy Change and Learning: An Advocacy Coalition Approach Boulder, CO: Westview Press. Sabatier, P. & Jenkins Smith, H. (1999). The advocacy coalition framework: an assessment. In: P. Sabatier, ed. Theories of the policy process Boulder, CO: Westview Press, 118 188. Sabatier, P.A. (1993) Policy Change over a Decade or More, in: Sabatier & Jenkins Smith (Eds), pp. 13 39. Sandberg, T ( 2012 ) State, Local Officials Clash Over Energy Development. The Observer, May 21, 2012. Retrieved from http://thecoloradoobserver. com/2012/05/state local officials clash over energy development in longmont/ Schattsc hneider, E. E. (1975). The Semi sovereign People (2nd ed., p. 143). New York: Wadsworth.

PAGE 235

222 Scholz, J. & N. Pinney. (1995). Duty, fear, and tax compliance: the heuristic bas is of citizenship behavior, American Journal of Political Science 39 : 490 512. Scholz, J. T. & Stiftel B. (eds). (2005). Adaptive Governance and Water Conflict: New institutions for collaborative planning Washington, DC: Resources for the Future. Schragg er, R. (2008). The progressive city. Paper from the Eleventh Annual Liman Colloquium at Yale Law School, 2008. Published by the Liman Public Interest Program at Yale Law School and the National State Attorneys General Program at Columbia Law School. SGEIS on the Oil, Gas and Solution Mining Regulatory Program. (2012). Retrieved June 26, 2012, from http://www.dec.ny.gov/energy/47554.html Shanahan, E. A ., Jones, M. D., & McBeth, M. K. (2011). Policy Na rratives and Policy Processes Policy Studies Journal, 39 (3), 535 561. doi:10.1111/j.1541 0072.2011.00420.x Shanahan, E. A ., McBeth, M. K., Hathaway, P. L., & Arnell, R. J. (2008). Conduit or contributor? The role of media in policy change theory. Policy S ciences, 41 (2), 115 138. doi:10.1007/s11077 008 9058 y Sheldon, E. (August 2, 2014). The Marijuana Divide. New York Times. http://www.nytimes.com/video/opinion/100000003035351/high time the marijuana divide.html?playlistId=1194811622182®ion=video grid&version=video grid thumbnail&contentCollection=Times+Video&contentPlacement=3 &module=recent videos&action=click&pgType=Multimedia&eventName=video grid click Shipan, C. R., & Volden, C. (2006). Bottom Up Federalism: The Diffusion of Antismoking Policies from U.S. Cities to States. American Journal of Political Science, 50 (4), 825 84 3. doi:10.1111/j.1540 5907.2006.00218.x Silverman, L. ( 2014 ) Denton City Council to Vote on Fracking Ban Tuesday. KERA News. July, 14, 204. Retrieved from StateImpact, A reporting project of NPR members station, https://stateimpact.npr.org/texas/2014/07/14/denton city council to vote on fracking ban tuesday/ Simon, H.A. (1957). Models of Man New York, Wiley & Sons. Smith A. (2000). Policy ne tworks and advocacy coalitions: explaining policy change and stability in UK industrial pollution policy? Environment and Planning C: Government and Policy 18 (1), 95 114 Snow, N. (November 21, 2013). House votes to restrict federal regulation of fracing. Oil and Gas Journal Retrieved from http://www.ogj.com/articles/2013/11/house votes to restrict federal regulation of fracing.html Stata Annotat ed Output Ordered Logistic Regression. (n.d.) UCLA: Statistical Consulting Group. from http://www.ats.ucla.edu/stat/stata/output/stata_ologit_output.htm (accessed September 4 2015).

PAGE 236

223 STATA ( n.d. ). MCA Multiple and joint correspondence analysis. Retrieved from http://www.stata.com/manuals13/mvmca.pdf STATA. (n .d. ). Title: mlogit postestimation: postesimation tools for mlogit. Published by STATA. Retrieved from www.stata.com/manuals13/rmlogitpostestimation.pdf State Land Oil and Gas Leasing (2012). New York: State Oil and Leasing. Retri eved June 26, 2012, from http://www.dec.ny.gov/energy/1528.html Stevens J P (1992) Applied multivariate statistics for the social sciences (2nd edition). Hillsdale, NJ: Erlbaum. STRONGER: State Review of Oil & Natural Gas Environmental Regulations. (2011). Colorado Hydraulic Fracturing State Review Oklahoma City, OK: STRONGERINC.ORG. STRONGER: State Review of Oil & Natural Gas Environmental Regulations. (2003). Texas State Review Oklahoma City, OK: STRONGERINC.ORG. Tarrow, S. (2013). Contentious Politics. The Wiley Blackwell Encyclopedia of Social and Political Movements. Teske, P. ( 2004 ) Regulations in the State Washington, DC: Brookings Institution Press. Teske, P., & Schneider, M. (1994). The Bureaucratic Entrepreneur: The Case of City Managers. Public Administration Revie w 54(4): 331 40. The State of Colorado, Official Site of Governor John Hickenlooper. (Aug 4, 2014) Gov. Hickenlooper announces task force to address local control and land use issues [Press Release]. Retrieved from https://www.colorado.gov/governor/news/gov hickenloop er announces task force address local control and land use issues 0 Thompson M., Ellis R. & Wildavsky A. (1990). Cultural Theory Westview Press, Boulder, Co. Tiebout, C. M. (1956). A Pure Theory of Local Expenditures. Journal of Political Economy, 64 (5), 416 424. Tilly, C. & Tarrow, S. ( 2007 ) Contentious Politics. Boulder, CO. Paradigm Publishers. Tocqueville, A., & Bender, T. (1981). Democracy in America. New York: Modern Library. Walker, J. L. (I991). Mobilizing Interest Groups in America Michigan: Ann Arbor. Walker, L. & Rice, K. (2011). Water Quality and the Interaction of Federal, State and Local Regulation of O i l and Gas Development in Colorado. Intermountain Oil and Gas BMP Project: Nature Resources Law Center. Weible, C. M. & Heikkil a, T. (2016). Comparing the Politics of Hydraulic Fracturing in New York, Colorado, and Texas. Review of Policy Research, 33 (3), pp. 232 250. doi. 10.1111/ropr.12170.

PAGE 237

224 Weible, C. M. (2006). An Advocacy Coalition Framework Approach to Stakeholder Analysis: U nderstanding the Political Context of California Marine Protected Area Policy. Journal of Public Administration Research and Theory, 17 (1), 95 117. doi:10.1093/jopart/muj015 Weible, C. M., & Sabatier, P. (2007). A guide to the advocacy coalition framework In the Handbook of the Public Policy Analysis by Fischer, F., Miller G.J., and M. S. Sidney. (eds). Taylor & Francis. Boca Raton, FL. Weible, C. M., & Sabatier, P. A. (2009). Coalitions, science, and belief change: comparing adversarial and collaborati ve policy subsystems. Policy Studies Journal, 37 (2), 195+. Retrieved from http://0 link.galegroup.com.skyline.ucdenver.edu/apps/doc/A20017 5347/UHIC?u=auraria_main &xid=3e836c83 Weible, C. M., & Elgin, D. (2013). Contrasting Capacities From City to International Levels of Government in Addressing Climate and Energy Issues. Cityscape Vol. 15, No. 1, Climate Change and City Hall (2013), pp. 163 179 Weible, C. M., P.A. Sabatier, & M. Lubell. (2004). A Comparison of a Collaborative and Top Down Approach to the Use of Science in Policy: Establishing Marine Protected Areas in California. Policy Studies Journal 32 (2): 187 208. Weible, C. M., Pierce J. J., & Heikkila, T. (2013). Explaining political activities in hydraulic fracturing in oil and gas development (pp. 1 25). Presented at the American Political Science Association Annual Conference on September 1, 2013. Weible, C. M., Sabatier, P., & Mc Queen, K. (2009). Themes and variations: Taking stock of the advocacy coalition framework. The Policy Studies Journal 37 (1), p. 121 140. Weible, C., Heikkila, T., DeLeon, P., & Sabatier, P. (2012). Understanding and influencing the policy process. Policy Sciences, 45 (1), 1 21. Retrieved from http://www.jstor.org/stable/41487060 Weimer, D L. (2008). Theories of and in the Policy Process. Policy Studies Journal 36 (4): 489 95. Wernerfelt, B. (1984). A resource based view of the firm. Strategic Management Journal, 5, 171 180. Stargazzette.com Retrieved June 26, 2012, from http://www.stargazette.com/article/20100201/NEWS01/2010363/Severance tax natural gas would offset cost 35 new regulatory jobs Wildavsky A. (1987). Choosing preferences by constructing institutions: a cultural theory of preference formation. American Political Science Review, 81 (1): 3 21. Williams, R. (2011). Lecture notes. Using the Margins Command to Estimate and Interpret Adjusted Predictions and Marginal Effects. Presented at Stata Conference on July 2011: Chicago, IL.

PAGE 238

225 Williams, R. ( 2012 ) Using the margins command to estimate and interpret adjusted predictions and marginal effects. The Stata Journal 12 (2), p. 308 331. Williams, R. ( 2017a ) Marginal effects for continuous variables. Richard Wil liams, University of Notre Dame, revised January 24, 2017.Accessed from http://www3.nd.edu/~rwilliam/xsoc73994/Margins02.pdf Williams, R. ( 2017b ) Understanding and interpreting the eff ects of continuous variables: the MCP (MarginsContPlot) Command. Richard Williams, University of Notre Dame, revised January 24, 2017. Accessed from http://www3.nd.edu/~rwilliam/xsoc73994 /Margins03.pdf Wilson, G. (2012). State and local regulation of oil & gas operations in Colorado. Knowledge Now. Colorado Municipal League. Worsham, J. (1997). Boulder, CO: Westview. Wright, J. R. (1996). Interest groups and congres s. Boston, MA: Allyn and Bacon Zafonte, M., and Sabatier, P. (1998). Shared beliefs and imposed interdependencies as determinants of ally networks in overlapping subsystems. Journal of Theoretical Politics 10 (4): 473 505.

PAGE 239

226 APPENDI X APPENDIX A Descriptive Statistics Preferred level of government A comparison of mean regulator score, where 1 = no regulator, 2 = local government, 3 = state government, and 4 = federal government, shows in general the issues identified as having smaller externalities, have lo wer preferred level of government scores. Table 1 : Preferred level of government for regulation by issue. Count, mean, and standard deviation shown. Preferred Regulator Issue No Regulation Local Gov. State Gov. Federal Gov. Mean S.D. Setbacks 3 98 105 22 2.64 0.67 Nuisances issues 2 121 85 19 2.53 0.66 Monitor water quality 1 31 138 58 3.11 0.63 Monitor air emissions 2 22 135 69 3.19 0.63 Core belief: Governmental ideology Table 2 : Governmental ideology, or core belief, of policy actors. The government should do more to advance society's goals, even if that means limiting the freedom and choices of individuals Government should put limits on the choices individuals can make so they do not get in the way of what is good for society Strongly Disagree Disagree Agree Strongly Agree Total Strongly Disagree 39 7 2 1 49 Disagree 13 35 10 0 58 Agree 6 19 38 4 67 Strongly Agree 1 1 4 10 16 Total 59 62 54 15 190 Pearson chi2(9) = 173.92 Pr = 0.000

PAGE 240

227 An iterated principle axes (ipf ) Factor analysis with varimax rotation was used to combine the scores for the two government attitude questions. The resulting factor loaded with an initial eigenvalue of 1.32, accounted for 100 percent of the variance, and had a uniqueness of .3405 for e ach initial variable. A varimax rotation resulted in no change in the single factor solution. The resulting single variable is continuous with a mean of 0 and ranging from 1.12 to 1.69. Policy change preference. Thirty eight percent of respondents indic ated they prefer fracking to be stopped or limited, 30 percent preferred fracking to continue at the current rate, and 32 percent preferred for fracking to expand. (TABLE Y). Table 3 : Policy change preference for hydraulic fracturing based oil and gas dev elopment by state and for all respondents. Position Colorado Texas Total Freq. Percent Freq. Percent Freq. Percent Stop/Limit 51 35% 34 45% 85 38% Continue at current rate 47 32% 20 26% 67 30% Expand 48 33% 22 29% 70 32% Total 146 76 222

PAGE 241

228 H 1 : T abular data Table 4 The instantaneous rate of change of governmental attitude on local and federal regulator preference for issues with broad and localized externalities. Issue Preference Externality dy/dx Std. Err. z P>z [95% Conf. Interval] Air Federal Preference Broad 0.082*** 0.028 2.97 0.003 0.028 0.137 Air Local Preference Broad 0.055*** 0.024 2.29 0.022 0.103 0.008 Water Federal Preference Broad 0.031 0.027 1.16 0.244 0.021 0.083 Water Local Preference Broad 0.065*** 0.029 2.26 0.024 0.121 0.009 Nuisance Federal Preference Localized 0.003 0.020 0.17 0.863 0.035 0.042 Nuisance Local Preference Localized 0.012 0.045 0.28 0.782 0.100 0.075 Setbacks Federal Preference Localized 0.007 0.021 0.33 0.743 0.035 0.049 Setbacks Local Preference Localized 0.005 0.039 0.12 0.904 0.072 0.081

PAGE 242

229 Margins For Interaction Between Policy Preference And State Table 5. Predictive Margins for Setbacks: Interaction between policy preference and State. Policy Preference with State Margin Std. Err. z P>z [90% Conf. Interval] Stop_Limit#Colorado 0.4945 0.0915 5.4 0.0000 0.3439 0.6450 Stop_Limit#Texas 0.7254 0.0736 9.86 0.0000 0.6044 0.8463 Continue#Colorado 0.2853 0.0666 4.28 0.0000 0.1757 0.3949 Continue#Texas 0.5808 0.1118 5.19 0.0000 0.3969 0.7647 Expand#Colorado 0.2251 0.0716 3.14 0.0020 0.1073 0.3428 Expand#Texas 0.5773 0.1192 4.84 0.0000 0.3812 0.7735 Table 6 Difference in margins between states (Texas Colorado): Setbacks Policy Preference Difference in Margins (Texas Colorado) Stop or limit 0.2309 Continue 0.2955 Expand 0.3523

PAGE 243

230 Table 7 Predictive Margins for Nuisance Issues: Interaction between policy preference and State. Predictive Margins (Delta Method) Policy Preference with State Margin Std. Err. z P>z [90% Conf. Interval] 1#Colorado 0.6672 0.0857 7.79 0.0000 0.5263 0.8082 1#Texas 0.6532 0.0829 7.88 0.0000 0.5169 0.7895 2#Colorado 0.4999 0.0829 6.03 0.0000 0.3635 0.6363 2#Texas 0.4784 0.1110 4.31 0.0000 0.2959 0.6609 3#Colorado 0.4701 0.0847 5.55 0.0000 0.3308 0.6093 3#Texas 0.4960 0.1115 4.45 0.0000 0.3126 0.6795 Table 8 Difference in margins between states (Texas Colorado): Nuisance issues. Policy Preference Difference in Margins (Texas Colorado) Stop or limit 0.0140 Continue 0.0215 Expand 0.0259

PAGE 244

231 Table 9 Predictive Margins for Air Emissions: Interaction between policy preference and State Policy Preference with state Margin Std. Err. z P>z [90% Conf. Interval] 1#Colorado 0.1659 0.0596 2.78 0.0050 0.0679 0.2639 1#Texas 0.1358 0.0632 2.15 0.0320 0.0318 0.2397 2#Colorado 0.0862 0.0458 1.88 0.0600 0.0108 0.1617 2#Texas 0.0728 0.0451 1.61 0.1060 0.0014 0.1469 3#Colorado 0.0494 0.0370 1.34 0.1820 0.0114 0.1103 3#Texas 0.0438 0.0376 1.17 0.2440 0.0180 0.1056 Table 10 Difference in margins between states (Texas Colorado): Air Emissions. Policy Preference Difference in Margins (Texas Colorado) Stop or limit 0.0301 Continue 0.0135 Expand 0.0056

PAGE 245

232 Table 11 Predictive Margins for Water Quality: Interaction between policy preference and State Policy Preference with state Margin Std. Err. z P>z [90% Conf. Interval] 1#Colorado 0.2102 0.0650 3.23 0.0010 0.1033 0.3171 1#Texas 0.2215 0.0792 2.8 0.0050 0.0912 0.3518 2#Colorado 0.0806 0.0414 1.95 0.0520 0.0125 0.1487 2#Texas 0.1097 0.0564 1.95 0.0520 0.0170 0.2025 3#Colorado 0.0990 0.0514 1.93 0.0540 0.0145 0.1835 3#Texas 0.1909 0.0978 1.95 0.0510 0.0301 0.3517 Table 12 Difference in margins between states (Texas Colorado): Water Quality. Policy Preference Difference in Margins (Texas Colorado) Stop or limit 0.0113 Continue 0.0291 Expand 0.0919

PAGE 246

233 APPENDIX B Appendix B includes additional information related to Chapter 3. Descriptive statistics for RQ1. Table 1 Summary of Number of Venues Shopped Number of Venues Shopped Frequency Percent 0 21 19.44 1 5 4.63 2 2 1.85 3 13 12.04 4 32 29.63 5 14 12.96 6 14 12.96 7 7 6.48 Table 2 Obs Mean Std. Dev. Min Max Average Resources 152 1 0.867193 0 2.833333 Table 3 Organization Type D istribution Group Type Frequency Percent Interest Group 100 65% Government 40 26% Academic/Consultant 14 9% Total 154 100%

PAGE 247

234 Table 4 Policy preference of respondents toward fracking. Fracking in the United States Frequency Percent Stopped or Limited 68 54% Continue at Current Rate 13 10% Expand Moderately or Extensively 45 36% Total 126 100% Table 5 Organizational Focus of Respondents. Focus of Organization Freq. Percent Local focus 59 38.8% State focus 74 49.7% National focus 19 12.5% Total 152 100% Descriptive statistics for RQ2 Table 6 Venue Shopping Frequency by Respondent Organization Type. Shopping Frequencies Federal Gov. State Agencies NY Gov.'s Office State Legislature State Courts Local Courts Local Gov. Never 54 30 29 31 78 86 34 Yearly 35 35 34 33 21 15 21 Monthly 14 36 34 35 7 4 39 Weekly 7 8 12 10 2 3 15 Total 110 109 109 109 108 108 109

PAGE 248

235 Table 7 Percent of respondents who shopped a venue at least once per year. average agreement and average perceived influence. Venue Percent shopping at least once per year Average agreement* Average perceived influence** NY Governor's Office 73% 0.83 1.64 State Agencies 72% 0.09 1.43 State Legislature 72% 0.14 1.13 Local Government 69% 0.22 1.23 State Courts 28% 0.03 1.26 Local Courts 20% 0.04 0.96 Agreement

PAGE 249

236 Local Level Action Table 8 Number of New York municipalities with a fracking ban or moratorium in 2011 and 2013. County 2011 2013 Grand Total 23 160 Albany 5 Broome 1 Cayuga 9 Chenago 1 Cortland 1 2 Delaware 2 Dutch 1 Erie 2 3 Fulton 1 Herkimer 5 Livingston 12 Monroe 6 Montgomery 4 Niagra 2 Oneida 2 25 Onondaga 4 13 Ontario 2 13 Orange 2 Ostego 6 12 Schenectady 1 Schohaire 8 Seneca 1 Steuben 1 Sullivan 5 Tioga 1 Tompkins 3 8 Ulster 7 Yates 3 7 Un identifiable 2 Source: Compiled by the author using information from Fracktracker.org

PAGE 250

237 Alternative Model for RQ1 with Holyoke et al. (2012), however their venue options did not include courts. Given that policy actor may only engage in a single court cases and expend large amounts of resources, while others may lobby a legislature at multiple hearings within the sam e time period, the model included in the Table 9 Ordered logit results for total venues shopped showing odds ratio Resources Org Type Org Focus Full Model Average Resources 2.280*** 2.349*** 2.052*** 1.824** (3.35) (3.45) (2.76) (2.24) Interest Group 2.048* 2.452** 3.124*** (1.95) (2.25) (2.72) Local Focus Comparison Organizational Focus State Focus 0.475* 0.480* ( 1.80) ( 1.74) National Focus 1.128 1.429 (0.20) (0.59) Belief: Stop/Limit Comparison Position on Fracking Belief: Continue 0.280** ( 1.97) Belief: Expand 0.348** ( 2.51) Observations 108 108 108 107 Pseudo R square 0.0311 0.0413 0.052 0.0741 Chi2 p value 0.0006 0.0004 0.0006 0.0001 Odds Ratios; Z statistic in parentheses. p<0.10, ** p<0.05, *** p<0.01.

PAGE 251

238 APPENDIX C Appendix C includes additional information related to Chapter 4. Respondent descriptive on policy position and org type Table 1 shows the average and standard deviation of the fracking policy preference score with the manually separated local government groups. The table is sorted by the mean score, showing the interest groups at the two extreme ends (1 = fracking should be stopped ; 5 = fracking should be expanded extensively). Table 1 Fracking policy preference score by organization affiliation. Organization affiliation Mean Std Dev. N Environmental groups 1.64 0.48 28 Local Gov., Anti Fracking 2.00 0.00 12 Federal Gov. 2.77 1.01 13 State Gov. 3.18 0.60 11 Academics and consultants 3.33 1.00 9 Local Gov., Pro Fracking 3.56 0.75 27 Oil and gas industry 3.87 0.80 39 Total 3.00 1.12 139 H1: ANOVA Table 2 Analysis of variance of policy preference and organization affiliation Number of obs. 139 R squared = 0.4826 Root MSE = 0.8203 Adj R squared = 0.4632 Source Partial SS df MS F Prob>F Model 83.492 5 16.698 24.81 0.000 affiliation 83.492 5 16.698 24.81 0.000 Residual 89.501 133 0.673 Total 172.99 138 1.254

PAGE 252

239 H1: Full Fisher Hayter Table Table 3 shows the fracking policy preference score comparison for each possible combination of organizational affiliation. Note that beyond local governments having statistically different and more moderate positions than interest groups, federal governments do a lso. Further, state government respondents had significantly different positions than environmental groups, but no difference when compared to oil and gas industry respondents. Likewise, the academic and consulting group only had statistically different sc ores when compared to the respondents associated with environmental groups. Table 3 Pairwise comparison of policy position scores using Fisher Hayter method. Group 1 vs Group 2 Mean 1 Mean 2 Difference FH Test Local Gov vs Oil and Gas 3.08 3.87 0.79 6.05* Local Gov vs Env. Groups 3.08 1.64 1.43 9.98* Local Gov vs Fed Gov 3.08 2.77 0.31 1.66 Local Gov vs State Gov 3.08 3.18 0.10 0.53 Local Gov vs Academics 3.08 3.33 0.26 1.20 Fed Gov vs State Gov 2.77 3.18 0.41 1.74 Fed Gov vs Oil and Gas 2.77 3.87 1.10 5.94* Fed Gov vs Env. Groups 2.77 1.64 1.13 5.79* Fed Gov vs Academics 2.77 3.33 0.56 2.24 State Gov vs Oil and Gas 3.18 3.87 0.69 3.48 State Gov vs Env. Groups 3.18 1.64 1.54 7.46* State Gov vs Academics 3.18 3.33 0.15 0.58 Oil and Gas vs Env. Groups 3.87 1.64 2.23 15.51* Oil and Gas vs Academics 3.87 3.33 0.54 2.51 Env. Groups vs Academics 1.64 3.33 1.69 7.61*

PAGE 253

240 H1: Correspondence Analysis Table 4 Correspondence analysis of policy preference and organization affiliation Number of obs 139.0 Pearson chi2(20) 100.0 Prob > chi2 0.0000 Total inertia 0.7196 Number of dim. 2 Expl. inertia (%) 96.5 Dimension Singular Value Principal Inertia Chi2 Percent Cumulative Percent dim 1 0.769589 0.592267 82.33 82.31 82.31 dim 2 0.319414 0.1020251 14.18 14.18 96.49 dim 3 0.140443 0.0197242 2.74 2.74 99.23 dim 4 0.074398 0.0055351 0.77 0.77 100 Total 0.7195514 100.02 100

PAGE 254

241 H2: Difference in Means and Extremism To determine if the local government pro fracking and anti fracking groups were more or less extreme than their interest group allies the oil and gas industry and environmental groups, respectively this paper compared a normalized problem perception scores of each group of respondents To normalize the score, this paper used the absolute score subtracted from the moderate position value (3). If the absolute value of the interest group score was greater tha n the absolute value of the local government score, a 1 was assigned, indicating the interest group was more extreme. For example, if respondents associated with environmental groups indicated on average, that the issue of Contamination of ground and sur face water supplies from chemicals in hydraulic fracturing fluids their average problem perception score was 4.2. Subtracting the moderate value (3) from the average score would give the environmental group a normalized prob lem perception of 1.2 (4.2 3). And, in this example, the respondents associated with local governments with an anti fracking problem, and their average problem perception score was a 3.1. Subtracting the moderate value (3) from and the average score would give anti problem perception of 0.1 (3.1 3). Then, the absolute value of each score is compared showing that the interest group (normalized value 1.2) is more extreme than the local government group (normalized value of 0.1). Each of the twenty problem perception scores were compared across respondents associated with the local governmen t s in the pro and anti fracking coalition and respondents

PAGE 255

242 from the respective interest group allies Then, the Bonferroni analysis was used to determine if the difference in average problem perception scores was significant. Table 5 shows whether the local government and respective interest group score was more extreme and indicates the statistical significance of the difference. Results show the average score of respondents associated with environmental groups were more extreme in 17 of the 20 issues than the average score of respondents associated with anti fracking local governments. B ut only one of those differences was statistically significant at a p value < 0.10. Similarly, the average problem perception scores of respondents associated with the oil and gas industry group were more extreme in 16 of the 20 issues than average scores of respondents associated with the pro fracking local gover nments but only two of those differences were statistically significant at a p value < 0.10.

PAGE 256

243 Table 5 Compar ison of mean problem perception scores between local governments and their interest group allies. Issue Statement Env. Group more extreme than Local Gov., Anti Industry group more extreme than Local Gov., Pro Misinformation among the general public about the risks, benefits, and effects of hydraulic fracturing. 1 1 Contamination of ground and surface water supplies from chemicals in hydraulic fracturing fluids. 1** 1 A patchwork of local regulations on hydraulic fracturing. 0 1 Conflict between mineral rights and property rights owners. 1 0 Contamination of ground water from methane migration. 1 1 Degradation of air quality from fugitive methane emissions. 1 1 Degradation of air quality from flares, diesel exhaust, and dust from well site operations. 1 1 Competition for available water supplies from hydraulic fracturing. 1 1 Nuisance to the general public caused by truck traffic, noise, and light from well site operations. 1 0 Surface degradation and erosion from access roads at well site operations. 1 1 Public distrust of the oil and gas industry. 0 1 Ineffective monitoring by state regulatory agencies of hydraulic fracturing. 1 1* Scare tactics and demonizing of the oil and gas industry by opponents of hydraulic fracturing. 1 1 Influence of the oil and gas industry over state administrative and legislative branches. 1 1** Boom and bust economic cycles from natural gas development. 1 1 Burdens on local government services from temporary employees for well site operations. 1 0 Risks of induced seismic activity caused by hydraulic fracturing. 0 1 Inadequate or incomplete communication by the oil and gas industry about the risks, benefits and effects of hydraulic fracturing to the general public. 1 0 Distribution of biased in formation against hydraulic fracturing. 1 1 Destruction of public lands by well site operations, processing facilities, and pipelines 1 1 Total count of interest group being more extreme than local gov. 17 16 p<0.10, ** p<0.05, *** p<0.01

PAGE 257

244 APPENDIX D Appendix D includes additional information related to Chapter 5. Factor Analyses of Resources Table 1 : Grouping resources into internal and external resource capacity Resource Mean Std. Deviation External (Factor 1) Internal (Factor 2) Access to elected political officials 2.02 0.95 0.818 0.104 Access to government officials 2.21 0.85 0.796 0.118 Access to media 2.01 0.88 0.743 0.272 Access to people with a similar position on hydraulic fracturing 2.31 0.87 0.720 0.280 Access to people with a different position on hydraulic fracturing 1.91 0.90 0.731 0.185 Support from the general public 1.56 0.86 0.531 0.408 Effective leadership in organization 2.13 0.91 0.419 0.502 Technical Support to Generate and disseminate information online 1.78 0.91 0.353 0.713 Financial resources 1.44 0.90 0.229 0.700 Generate and disseminate scientific reports and analysis 1.51 0.96 0.026 0.829 Extraction Method: Principal Component Analysis. Rotation Method: Varimax with Kaiser Normalization. Cronbach's Alpha: Factor 1 = 0.875; Factor 2 = 0.733

PAGE 258

245 Table 2 : Factor Analysis of Political Activities Activity Mean Std. Deviation Primary (Factor 1) Secondary (Factor 2) Participating in public meetings 2.53 1.44 0.842 0.213 Forming and maintaining a coalition with allies 2.49 1.86 0.800 0.336 Lobbying elected officials 1.81 1.85 0.788 0.215 Testifying at public hearings 1.55 1.42 0.707 0.364 Communicating with the news media 2.12 1.78 0.680 0.444 Posting information or advocating online 2.10 1.93 0.413 0.695 Generating and disseminating research and reports 1.79 1.61 0.362 0.568 Formal complaining to regulatory commissions 1.03 1.36 0.259 0.688 Taking legal action (e.g., lawsuits) 0.46 0.91 0.259 0.679 Organizing or Participating in public protests 0.39 0.89 0.151 0.809 Extraction Method: Principal Component Analysis. Rotation Method: Varimax with Kaiser Normalization.

PAGE 259

246 H1: Comparison of R esource s External Resources Table 3a ANOVA of External Resources by Organization Affiliation Number of obs 119 R squared = 0.1027 Root MSE 0.558263 Adj R squared = 0.063 Source Partial SS df MS F Prob>F Model 4.0324308 5 0.80648616 2.59 0.0296 affiliation 4.0324308 5 0.80648616 2.59 0.0296 Residual 35.217335 113 0.31165783 Total 39.249766 118 0.33262514 Table 3b Fisher Hayter pairwise comparisons of external resources Group 1 vs Group 2 Group 1 Mean Group 2 Mean dif FH test Local Gov vs Fed Gov 1.69 1.47 0.21 1.61 Local Gov vs State Gov 1.69 1.67 0.02 0.13 Local Gov vs Oil and Gas 1.69 1.67 0.02 0.20 Local Gov vs Env. Groups 1.69 1.92 0.23 2.20 Local Gov vs Academics 1.69 1.19 0.50 3.39 Fed Gov vs State Gov 1.47 1.67 0.19 1.12 Fed Gov vs Oil and Gas 1.47 1.67 0.19 1.46 Fed Gov vs Env. Groups 1.47 1.92 0.45 3.19 Fed Gov vs Academics 1.47 1.19 0.29 1.65 State Gov vs Oil and Gas 1.67 1.67 0.00 0.00 State Gov vs Env. Groups 1.67 1.92 0.25 1.63 State Gov vs Academics 1.67 1.19 0.48 2.59 Oil and Gas vs Env. Groups 1.67 1.92 0.25 2.35 Oil and Gas vs Academics 1.67 1.19 0.48 3.23 Env. Groups vs Academics 1.92 1.19 0.74 4.74* p value < 0.10, ** p value < 0.05, *** p value < 0.01

PAGE 260

247 Internal Resources Table 4a ANOVA of Internal Resources by Organization Affiliation Number of obs 118 R squared = 0.1163 Root MSE 0.66963 Adj R squared = 0.0768 Source Partial SS df MS F Prob>F Model 6.6075893 5 1.3215179 2.95 0.0154 affiliation 6.6075893 5 1.3215179 2.95 0.0154 Residual 50.22133 112 0.44840473 Total 56.828919 117 0.48571726 Table 4b Fisher H ayter pairwise comparisons for internal resources. Group 1 vs Group 2 Group 1 Mean Group 2 Mean dif FH test Local Gov vs Fed Gov 1.36 1.67 0.31 1.95 Local Gov vs State Gov 1.36 1.69 0.34 1.90 Local Gov vs Oil and Gas 1.36 1.91 0.55 4.68* Local Gov vs Env. Groups 1.36 1.83 0.47 3.65* Local Gov vs Academics 1.36 2.03 0.67 3.77* Fed Gov vs State Gov 1.67 1.69 0.03 0.13 Fed Gov vs Oil and Gas 1.67 1.91 0.24 1.49 Fed Gov vs Env. Groups 1.67 1.83 0.16 0.95 Fed Gov vs Academics 1.67 2.03 0.36 1.73 State Gov vs Oil and Gas 1.69 1.91 0.21 1.19 State Gov vs Env. Groups 1.69 1.83 0.13 0.71 State Gov vs Academics 1.69 2.03 0.33 1.49 Oil and Gas vs Env. Groups 1.91 1.83 0.08 0.62 Oil and Gas vs Academics 1.91 2.03 0.12 0.68 Env. Groups vs Academics 1.83 2.03 0.20 1.08 p value < 0.10, ** p value < 0.05, *** p value < 0.01

PAGE 261

248 H2 : Primary and Secondary Activities Table 5. OLS analysis of the relationship between organization type and political activities, standard errors in parentheses. Primary Activities Secondary Activities Org. Type Local Gov Comparison Group Fed Gov 0.155 0.192 (0.430) (0.340) State Gov 0.339 0.391 (0.490) (0.380) Oil and Gas 0.685** 0.146 (0.320) (0.250) Env. Groups 0.660* 1.195*** (0.380) (0.300) Academics 0.681 0.21 (0.540) (0.390) Resource Capacity External Resource Capacity 0.666*** 0.234 (0.240) (0.190) Internal Resource Capacity 0.575*** 0.318* (0.210) (0.170) Years Involved 0 1 years Comparison Group 2 4 years 0.551 0.001 (0.510) (0.400) 5 9 years 0.066 0.044 (0.520) (0.410) 10 20 years 0.002 0.487 (0.530) (0.420) 21+ years 0.128 0.397 (0.550) (0.440) Position Anti fracking Comparison Group Pro fracking 0.444 0.09 (0.360) (0.280) Constant 0.143 0.119 (0.610) (0.480) R 2 0.479 0.44 p<0.10, ** p<0.05, *** p <0.01

PAGE 262

249 H2 : OLS Progression of primary and secondary activities. Table 6 Primary Activity Regression Progression, standard errors in parentheses. Base Model w/ year w/ position full Primary Activities Primary Activities Primary Activities Primary Activities Capacity Avg. External Capacity 0.657*** 0.655*** 0.688*** 0.666*** (0.230) (0.240) (0.230) (0.240) Avg. Internal Capacity 0.480** 0.567*** 0.459** 0.575*** (0.200) (0.210) (0.200) (0.210) Org. Type Local Gov Comparison Category Fed Gov 0.099 0.017 0.081 0.155 (0.390) (0.410) (0.410) (0.430) State Gov 0.312 0.155 0.354 0.339 (0.440) (0.470) (0.440) (0.490) Oil and Gas 0.769*** 0.545* 0.841*** 0.685** (0.290) (0.310) (0.310) (0.320) Env. Groups 0.985*** 0.942*** 0.878** 0.660* (0.300) (0.300) (0.370) (0.380) Academics 0.77 0.675 0.775 0.681 (0.500) (0.510) (0.530) (0.540) Years Involved 0 1 years Comparison Category 2 4 years 0.541 0.551 (0.510) (0.510) 5 9 years 0.145 0.066 (0.510) (0.520) 10 20 years 0.016 0.002 (0.530) (0.530) 21+ years 0.063 0.128 (0.550) (0.550) Position Anti fracking Comparison Category Pro fracking 0.171 0.444 (0.320) (0.360) Constant 0.215 0.085 0.129 0.143 (0.370) (0.570) (0.440) (0.610) R sqr 0.453 0.473 0.453 0.479 dfres 97 89 94 86 BIC 335.1 333.8 332.5 330.2 p<0.10, ** p<0.05, *** p<0.01

PAGE 263

250 Table 7 Secondary Activity Regression Progression Base Model w/ year w/position full Secondary Activities Secondary Activities Secondary Activities Secondary Activities b/se b/se b/se b/se Capacity Avg. External Capacity 0.219 0.23 0.233 0.234 (0.170) (0.190) (0.180) (0.190) Avg. Internal Capacity 0.298* 0.318** 0.290* 0.318* (0.150) (0.160) (0.160) (0.170) Org. Type Local Gov Comparison Category Fed Gov 0.185 0.255 0.106 0.192 (0.300) (0.320) (0.320) (0.340) State Gov 0.412 0.356 0.42 0.391 (0.360) (0.360) (0.370) (0.380) Oil and Gas 0.263 0.115 0.272 0.146 (0.220) (0.240) (0.240) (0.250) Env. Groups 1.261*** 1.254*** 1.253*** 1.195*** (0.230) (0.230) (0.290) (0.300) Academics 0.28 0.231 0.252 0.21 (0.360) (0.370) (0.380) (0.390) Years Involved 0 1 years Comparison Category 2 4 years 0.004 0.001 (0.390) (0.400) 5 9 years 0.032 0.044 (0.400) (0.410) 10 20 years 0.496 0.487 (0.410) (0.420) 21+ years 0.39 0.397 (0.430) (0.440) Position Anti fracking Comparison Category Pro fracking 0.013 0.09 (0.250) (0.280) Constant 0.004 0.165 0.008 0.119 (0.280) (0.440) (0.340) (0.480) R sqr 0.418 0.448 0.411 0.44 dfres 98 91 95 88 BIC 281.7 287.3 282.7 288.5 p<0.10, ** p<0.05, *** p<0.01

PAGE 264

251 Network Size Table 8 Network Size Regression Progression Base With Year With Capacity Full Network Size Network Size Network Size Network Size b/se b/se b/se b/se Capacity Avg. External Capacity 2.493*** 2.393*** (0.59) (0.60) Avg. Internal Capacity 0.826* 0.881* (0.48) (0.49) Org. Type Local Gov Comparison Category Fed Gov 0.658 1.614 1.666* 1.872* (1.19) (1.17) (0.94) (0.99) State Gov 0.508 1.392 1.907* 2.109* (1.30) (1.26) (1.03) (1.07) Oil and Gas 1.572* 1.841** 1.550** 1.744** (0.86) (0.84) (0.75) (0.77) Env. Groups 1.836* 2.234** 1.404* 0.959 (0.94) (0.89) (0.73) (0.91) Academics 0.288 0.421 1.073 1.492 (1.26) (1.19) (1.15) (1.18) Years Involved 0 1 years Comparison Category 2 4 years 0.161 1.513 1.421 (1.62) (1.26) (1.26) 5 9 years 0.874 0.895 0.827 (1.63) (1.26) (1.27) 10 20 years 0.356 0.496 0.478 (1.70) (1.32) (1.32) 21+ years 0.344 0.332 0.28 (1.77) (1.36) (1.37) Position Anti fracking Comparison Category Pro fracking 0.665 (0.83) Constant 5.128*** 5.859*** 1.047 1.528 (0.61) (1.59) (1.40) (1.49) R square 0.039 0.086 0.411 0.409 Degrees of freedom 138 116 101 98 BIC 817.6 699.3 576.6 570 p< 0.10, ** p<0.05, *** p<0.01

PAGE 265

252 Network pattern MCA MCA on network pattern Table 10a Multiple/Joint correspondence analysis Number of obs 126 Total inertia 0.1220 Number of axes 3 Method: Burt/adjusted inertias Table 10b Dimension Principal inertia percent Cumulative percent dim 1 0.0846201 69.37 69.37 dim 2 0.0138374 11.34 80.71 dim 3 0.0051033 4.18 84.9 dim 4 0.0001577 0.13 85.03 dim 5 0.0000327 0.03 85.05 dim 6 0.0000146 0.01 85.06 Total 0.1219846 100