Citation
Deconstructing political behavior : the influence of context and individual characters

Material Information

Title:
Deconstructing political behavior : the influence of context and individual characters
Creator:
Olofsson, Kristin L.
Place of Publication:
Denver, CO
Publisher:
University of Colorado Denver
Publication Date:
Language:
English

Thesis/Dissertation Information

Degree:
Doctorate ( Doctor of philosophy)
Degree Grantor:
University of Colorado Denver
Degree Divisions:
School of Public Affairs, CU Denver
Degree Disciplines:
Public affairs
Committee Chair:
Weible, Christopher M.
Committee Members:
Heikkila, Tanya
Swann, William
Nohrstedt, Daniel

Notes

Abstract:
People, policy, and power are inextricably linked. There is wide variation in how individuals access and are able to influence policy, which is disheartening for the practice of democracy and fair and equal representation. This research studies the execution of power within policy by exploring how people participate in policymaking, as evidenced through their political behavior. Political behavior comes in many forms, such as writing a letter to the editor, protesting, or lobbying. Current research might be able to tell us who is likely to participate or even when they might participate but it does not explain how that individual made his or her participation decision and in what ways that decision was enabled or constrained. This research explores the determinants of political behavior decisions by focusing on two aspects: context and individual characteristics. Taking an institutional view of decision-making, the research argues that external context, recognized here as sectoral affiliation, and internal mechanisms, captured through individual characteristics, interact in detectable and predictable ways to manifest in patterned political behavior. Results indicate three main themes. First, there is polarization in belief extremity related to perceptions of issue contentiousness. As an individual perceives an issue to be increasingly more contentious, they are also likely to associate heightened feelings of threat and a lack of viable solutions. Second, political behavior choices are not random; there is an element of intentionality or strategy behind the choices. Finally, context appears to amplify some individual characteristics and dampen others. The research bolsters explanatory capacity of theories of policymaking and contributes to practical knowledge about how we can better understand our setting in order to improve the quality of political discourse.

Record Information

Source Institution:
University of Colorado Denver
Holding Location:
Auraria Library
Rights Management:
Copyright Kristin Olofsson. Permission granted to University of Colorado Denver to digitize and display this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.

Downloads

This item has the following downloads:


Full Text
DECONSTRUCTING POLITICAL BEHAVIOR: THE INFLUENCE OF CONTEXT AND
INDIVIDUAL CHARACTERISTICS by
KRISTIN L. OLOFSSON B.A., University of Colorado, 2001 M.Sc., Uppsala University, 2012
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
2019


©2019
KRISTIN L. OLOFSSON
ALL RIGHTS RESERVED


This dissertation for the Doctor of Philosophy degree by
Kristin L. Olofsson has been approved by Public Affairs Program by
Christopher M. Weible, Chair Tanya Heikkila William Swann Daniel Nohrstedt
Date: May 18, 2019


Olofsson, Kristin L. (PhD, Public Affairs)
Deconstructing Political Behavior: The Influence of Context and Individual Characteristics Thesis directed by Professor Christopher M. Weible
ABSTRACT
People, policy, and power are inextricably linked. There is wide variation in how individuals access and are able to influence policy, which is disheartening for the practice of democracy and fair and equal representation. This research studies the execution of power within policy by exploring how people participate in policymaking, as evidenced through their political behavior. Political behavior comes in many forms, such as writing a letter to the editor, protesting, or lobbying. Current research might be able to tell us who is likely to participate or even when they might participate but it does not explain how that individual made his or her participation decision and in what ways that decision was enabled or constrained. This research explores the determinants of political behavior decisions by focusing on two aspects: context and individual characteristics. Taking an institutional view of decision-making, the research argues that external context, recognized here as sectoral affiliation, and internal mechanisms, captured through individual characteristics, interact in detectable and predictable ways to manifest in patterned political behavior. Results indicate three main themes. First, there is polarization in belief extremity related to perceptions of issue contentiousness. As an individual perceives an issue to be increasingly more contentious, they are also likely to associate heightened feelings of threat and a lack of viable solutions. Second, political behavior choices are not random; there is an element of intendonality or strategy behind the choices. Finally, context appears to amplify some individual characteristics and dampen others. The research bolsters explanatory capacity of
IV


theories of policymaking and contributes to practical knowledge about how we can better understand our setting in order to improve the quality of political discourse.
The form and context of this abstract are approved. I recommend its publication.
Approved: Christopher M. Weible
v


DEDICATION
This dissertation is dedicated to all the patience and hard work of my advisors, family, and
friends.
To Chris, Tanya, Will, Daniel, and many other mentors, your guidance throughout this process has been invaluable. I will be forever grateful for everything that you have done, and I hope to
continue the tradition with my own students.
To my parents Mike and Sylvia, you were my first and best teachers. You taught me everything
important.
To Koren and MJ, thank you for putting up with me and always being there, without judgement but usually with some sarcasm. I need someone to keep me in line.
To Rolf and Ulla, your support and curiosity has been such an encouragement throughout this process. You are always there with a smile and a willingness to listen, for which I cannot thank
you enough.
To my amazing children Emmelina and Stellan, you are my source of inspiration. Having you has made every bit of this better. I love you more than you will ever know.
And finally, to my amazing husband Robert, words cannot express how much you mean to me. You’ve supported me through it all, been there through the good and bad, and never wavered. You are my best friend and my favorite person. You really deserve all the credit.
VI


TABLE OF CONTENTS
I. INTRODUCTION......................................................................1
Background Literature on Political Behavior.......................................7
Conceptual Framework.............................................................12
Objective 1: Role of Individual Characteristics................................15
Objective 2: Activities........................................................15
Objective 3: Policy Networks...................................................17
Research Design..................................................................22
Case Selection.................................................................23
Methods of Data Collection.....................................................25
Sampling: Policy Actor Identification........................................25
Key Variable Operationalization..............................................28
Analysis.........................................................................29
Limitations......................................................................31
Contributions....................................................................32
II. PERCEPTIONS OF CONTENTIOUSNESS: HOW INDIVIDUAL TRAITS SHAPE
ENVIRONMENTAL POLICY CONFLICTS......................................................34
Case Setting: Hydraulic Fracturing...............................................36
Data Collection..................................................................38
Results..........................................................................40
Respondent Belief Structures...................................................41
vii


Conclusion
48
III. A CONFLICT WILD: EXPLAINING ACTIVITIES OF CONFLICT EXPANSION AND
CONTAINMENT.......................................................................52
Theoretical Background.........................................................54
Conceptualizing Political Participation: Expanding & Containing Conflict.....55
Explaining Political Participation: Sector and Individual Influences.........58
Sector.....................................................................58
Individual Beliefs.........................................................61
Model Specification............................................................64
Dependent Variable...........................................................65
Independent Variables........................................................69
Results........................................................................70
OLS Regression Models........................................................72
Component Analysis...........................................................75
Conclusion.....................................................................77
IV. STRATEGIC MOTIVATIONS IN POLICY GOVERNANCE....................................82
Literature Review..............................................................83
Conceptual Framework...........................................................87
Methodology....................................................................93
Results........................................................................96
viii
Exponential Random Graph Model
102


Conclusion...................................................................106
V. CONCLUSION..................................................................Ill
What We Learned..............................................................115
Limitations..................................................................117
Why It Matters...............................................................119
REFERENCES.....................................................................120
APPENDIX
A: Concepts and Operationalization..............................................135
B: Full Survey..................................................................136
C: Factor Analysis (Principle Components with Orthogonal Varimax Rotation)......144
D: Basic Statistical Description of Dataset.....................................145
IX


Chapter I
Introduction
An essential part of the policy process is the maneuvering of political behavior to impact policy change and outcomes. Political behavior comes in many different forms, such as lobbying, writing letters to the editor, or sharing your opinion with government agencies. The development of a special task force in the state of Colorado to deal with policy issues related to oil and gas regulations illustrates many aspects of political behavior. While Colorado has been involved in oil and gas extraction for many years, the issue has historically been contentious, particularly in recent years with the increased use of hydraulic fracturing as an extractive technique. As this highly-contentious issue has progressed, many individuals have been involved in a myriad of ways in attempting to influence the policy surrounding hydraulic fracturing. Some of the ways in which individuals have attempted to influence policy are creating citizen groups, joining environmental groups, testifying at public hearings, advocating with influential decision makers, and sharing their opinion in the media, among several others.
One example of a particularly visible way of participating in the policymaking process was the creation and implementation of a task force by then-Governor John Hickenlooper and former House Representative Jared Polis (now governor) to address a volatile situation that arose in 2014. Stemming from 2012 when the city of Longmont in Northern Colorado passed a ban on hydraulic fracturing, the task force was essentially a compromise between existing pro- and anticoalitions to avoid a ballot issue and a contentious lawsuit involving the city of Longmont and the Colorado State regulatory authority, the Colorado Oil and Gas Conservation Commission (COGCC) (Jaffe, 2014).
1


The task force was comprised of an assortment of representatives from industry, local government, and environmental and other interests. It was one of the first collaborative, inter -coalitional venues available in the highly conflictual policy issue of hydraulic fracturing in the state of Colorado.1 The creation of an inter-coalitional task force triggered a movement towards a less adversarial strategy from both coalitions. In this highly visible venue, individuals worked together to respond to a shifting landscape of public opinion and demands for transparency and accountability from all sides.
The actions of the individuals that brought about the creation of the task force varied, including but not limited to media campaigns, research production and dissemination, lobbying, and testifying at public hearings. There were numerous venues in which these actions occurred, and the networks of individuals involved were diverse, from federal to local-level government officials to community nonprofits to the general public. The coalescence of activities and policy networks into political behavior prompted Governor Hickenlooper to respond, in what eventually had a sizeable impact on the November 2014 ballot issues in Colorado.
Current theory in political behavior might be able to predict when an individual participates and perhaps even why that individual chooses to participate, but existing theory does not tell us how an individual made that political behavior decision. We cannot understand the impact of political behavior within the policy process without understanding how individuals make decisions. To better understand individual political behavior within the policy process, this dissertation focuses on two of its aspects: activities and policy networks. Activities are defined as the actions taken by policy actors to influence the policy process, such as the lobbying efforts of
1 One of the immediate outcomes of the creation of the task force is that four citizens’ initiatives related to oil and gas issues slated for the November 2014 ballot were dropped. Two initiatives were from the anti-drilling coalition and two were from the oil and gas industry and its allies.
2


the individuals involved in the Colorado case. Policy networks represent the interactions of individuals and organizations with an aim to achieve personal or professional goals. This political behavior is influenced by two general factors: context and individual characteristics. Institutional context is comprised of the rules and norms of the broader setting in which an individual is operating, such as the public, private, or nonprofit sector. The institutional context varies the rules of the game in different settings. Individual characteristics are defined as personal belief systems or cognitive tendencies, such as the tendency to hold strong beliefs regarding a policy issue or an individual’s risk perceptions or determination of issue salience. Figure 1 displays the relationship between these concepts, ultimately producing political behavior.
POLITICAL BEHAVIOR
Figure 1: Model of Political Behavior
This dissertation uses a multi-method case approach to explain how institutional context and individual characteristics interact to modify political behavior in a high-conflict policy issue. This research abandons when questions of political behavior and instead explores how and why institutional conditions and individual characteristics interact to constrain or enable that behavior by asking: How do institutional context and individual characteristics influence political
3


behavior in contentious policy debates? The dissertation views political behavior through the theoretical lens of the Advocacy Coalition Framework (ACF) and supplements from several theories such as institutionalism, policy networks, conflict expansion, and behavioral economics to develop three chapters that will explore different aspects of how context modifies political behavior.
Chapter II: Role of Individual Characteristics. Objective: To describe how individual characteristics shape environmental policy conflicts. Understanding the policy process requires understanding the individuals involved in environmental policy making. This objective will consider how the beliefs and policy preferences of individuals involved in the policies shape the extraction of unconventional oil and gas in the US. Data for this foundational chapter is drawn from a 2016 survey and associated interviews of individuals involved in hydraulic fracturing at the national level.
Chapter III: Activities. Objective: To identify patterns in the types of activities used by policy actors in the policy process and how these patterns are influenced by institutional context and individual characteristics. According to the ACF, individuals participate in the policy process, via coalitions, with intent to influence the policy process (Sabatier & Jenkins-Smith, 1993). This objective explores how institutional context, such as sector, may interact with individual-level characteristics, such as strength of beliefs, to encourage or thwart individual activity within the policy process. This chapter dives into the context of Colorado, using data from a set of surveys and associated interviews of individuals involved in hydraulic fracturing in the state. The survey was first done in 2015 and repeated in 2017.
Chapter IV: Networks. Objective: To identify variation in policy networks and how this variation can be explained by differences in institutional context and individual characteristics.
4


The ACF purports that policy issues are characterized by competing advocacy coalitions, which can be understood as networks of individuals (Matti & Sandstrom, 2013). This objective employs a relational approach (Ingold, 2011) to explore the policy networks of individuals involved in unconventional oil and gas development in Colorado. While the ACF recognizes the fundamental role of individual choice in network behavior, it lacks theory regarding how those network choices might be influenced by institutional context and individual characteristics. This objective applies theories of behavioral economics and strategic network formation to assess the self-interested and benefit-maximizing choices of individuals on network structure, using the same Colorado dataset as in the previous chapter.
By focusing on the influences of political behavior decisions, the dissertation generates knowledge about the diversity of political behavior that might be used in different contexts and contributes to our understanding of the policy process and political outcomes. The theoretical definition of political behavior can be refined by exploring the use of activities and policy networks at an individual level. Exploring distinct institutional settings with varying options for activities and political networks can lend insight into why certain political behaviors emerge. Individuals filter institutional context through their individual characteristics, and that influences their political behavior. We can detect institutional settings and individual cognitive characteristics that shape political behavior. In addition, this research has implications for understanding contentious political exchanges. If we better understand the different toolbox that each person has to interact in the policy process, then we can tackle difficult issues such as disproportionate representation and unequal and inequitable outcomes. Democracy is built upon a fundamental tenet that each person - if they so desire - should have a voice in the process. The
5


dissertation proposed here aims to clarify how people choose to exercise that voice and what determines those individual decisions.
The setting for the research is the exploration of shale oil and gas through hydraulic fracturing. Hydraulic fracturing is a technique used to extract natural gas and oil from unconventional sources. To reach unconventional sources of oil and natural gas, vertical wells are first drilled and then a horizontal well is drilled to the source point. Hydraulic fracturing fluids are forced into the wells at high pressure in order to fracture the rock and prop open the fractures after breakage. The released resource can then flow through the horizontal and then the vertical wells for extraction. With advancements in horizontal drilling, the technique has grown markedly in its usage in the past two decades, which has been the source of extensive debate in the United States and worldwide (Heikkila, Pierce, et al., 2014; Rahm, 2011). Issues surrounding hydraulic fracturing include air and water quality, public nuisance, energy security, usage of public and private land, mineral rights, siting of wells, and impacts on the economy.
The research first approaches political behavior within the policy process by exploring the general associations between individual characteristics and policy conflicts. This work is done at the national scale, to avoid possible issues of contextual variation. However, the primary interest of the dissertation is precisely contextual variation, which is addressed through the case study in Colorado. The state is a representative setting in which to explore political behavior within the policy process as the state has a long history with oil and gas development and is accustomed to issues associated with this particular industry, while opposition remains visible and vocal with a diversity of actors in a diversity of institutional settings. The study population for this research is not the public but rather individuals who are already involved in the policy process. These individuals are known as policy actors. In the earlier example, policy actors were
6


those individuals who were involved in the debates that led to the creation of the task force. Drawing upon surveys and interviews in Colorado in 2015, and 2017, policy actors were asked how they participate in the policy process and to what extent and toward whom they direct their activities. Although the survey sample was directed towards policy actors ostensibly already involved in the policy process, or at least professionally attached to the issue, there was wide variation in levels of participation among these individuals, as well as in institutional context and individual characteristics.
Background Literature on Political Behavior
Participation in the policy process is a vital part of a functioning democracy. In an open, transparent democracy, political behavior is crucial to claims of representativeness and accountability, particularly in a pluralistic political system such as the United States. Public policy, as defined by Dye, encompasses both government action and inaction - what governments choose to do as well as not to do (2002). Participation in the public policy process can provide input for decision makers; it can bring about learning and new information - perhaps even policy change - and is fundamental to democratic norms (Nohrstedt, 2013; O’Faircheallaigh, 2010).
There has been extensive work on political participation generally (Brady, Verba, & Lehman Schlozman, 1995; Hillman & Hitt, 1999; Kaase, 2009; Milbrath, 1965; Mutz, 2002; Smith, 1984; Verba & Nie, 1987; Webler, Tuler, & Krueger, 2001). Milbrath put forth an operational definition of political participation as voting, campaign activity, community participation, and particularized contacting (1965). There has also been considerable theorizing about both the determinants of and outcomes of participation. Early empirical work on political participation was dominated by voting studies as researchers explored what motivates citizens to
7


vote, and this continues to comprise the bulk of work on political behavior (Kaase, 2009). Unconventional forms of political participation, such as protesting, are gaining ground in the empirical literature but nonetheless, the focus remains on convention political participation as advanced studies unpack the multi-level and interrelated determinants of participation.
The current literature regarding political behavior is limited in three major ways. First, knowledge surrounding political behavior narrowly focuses on elections and legislative issues. The policy process is much broader and can encompass everything from the antecedents of policy change through implementation. Second, theorization and empirical work in political behavior tends to focus on the general public and ignores elites and other interested parties. Finally and relatedly, the concept of political behavior is often limited to participation such as voting. This is only one aspect of political behavior. Some of the foundational ideas regarding political behavior - most often participation - within the policy process literature came from Kingdon, who drew upon participation as a means to illuminate the role of solution development in explaining policy change (Kingdon, 1984). Nearly ten years later, Roberts and King used grounded theory to explore how highly active individuals in the policy process might decide whether to participate, and if so, how (1991). The work of these scholars gave some clarity about the determinants of participation and how participation might contribute to outcomes.
Bishop and Davis undertook a thorough review of the extant literature surrounding political behavior in the policy process and concluded that the typologies and frameworks remain disjointed and in need of empirical testing (2002). They define participation as “the expectation that citizens have a voice in policy choices” (ibid.: 14). The authors assert that participation happens as the result of an understanding of shared power between citizens and the government. For Bishop and Davis, participation is possible only when citizens believe that their actions will
8


impact government policy. From this lens, the authors divide the field of policy participation studies into approaches along a continuum of realized citizen expectations. While these approaches all have positive contributions to understanding participation in the policy process, Bishop and Davis ultimately argue that theorizing about participation is “laden by idealist notions of democracy” (2002: 14) and would benefit from clarity gained through empirical research.
More ideas about political behavior within the policy process can be found in the major theories of the policy process. Schattschneider (1960), Lowi (1972), and Wilson (1973) all recognized the iterative cycle between policymaking and individual political behavior. The policy feedback literature explores how a policy can impact civic engagement (Mettler, 2002; Pierson, 1996) by acknowledging that political actors “respond to policies after enactment just as much as they would have before passage... for reasons that are exogenous to the policy itself’ (Moynihan & Soss, 2014: 321). Punctuated equilibrium theory (PET) rests upon the assumption that individuals participate in the policy process in a myriad of ways that ultimately push policy making from maintenance of the status quo through incremental changes to major shifts in policy (Baumgartner & Jones, 1993; B. D. Jones & Baumgartner, 2012). In a massive empirical effort, Baumgartner et al. interviewed 315 individuals involved with lobbying in Washington, D.C. and analyzed hundreds of thousands of contextual documents to explore the relationship between lobbying and policy change (Baumgartner, Berry, Hojnacki, Kimball, & Leech, 2009). While demonstrating empirically the theoretical conclusions of PET, the authors also explored how interest group arguments are shaped and the tactics that advocates employ.
Schneider and Ingraham’s theory of social construction (1993) argues that policies are shaped by the political behavior of policy actors, and vice versa. The theory asserts that political
9


behavior can be socially constructed within the policy process as target populations translate signals regarding their societal status from enacted policies and policy makers. Fundamentally, social construction theory argues that “the social construction of target populations has a powerful influence on public officials and shapes both the policy agenda and the actual design of policy” (ibid.: 334). Political behavior is motivated by the messages that we receive regarding target populations. While Schneider and Ingraham’s work and the work of other policy process theorists has lent some clarity to political behavior, additional empirical and theoretical knowledge would contribute to a better understanding of this mechanism within the policy process.
Scholars appreciate that we need to increase empirical and theoretical awareness of how individuals participate in the policy process. Rydin and Pennington agree that theory regarding political behavior should be grounded by exploring the incentive structures of the involved individuals to ascertain how those individuals choose to participate, and what the resulting outcomes might be (2011). Individual political behavior varies across institutional contexts, and there is a lack of empirical research that identifies how different institutional contexts affect behavior (Nohrstedt, 2013). Zaller developed a model of public opinion formation that posited that citizens receive cues from policy elites and form opinions from which they are unlikely to deviate (1992). Termed the “elite-leadership model of public opinion” (Kam, 2012), Zaller’s model of public political behavior explains the role of intermittent information processing. Zaller’s foundational work has been highly influential in developing a research agenda on public political behavior that has produced a wealth of empirical findings and theoretical insights (see Entman, 1993; Inglehart & Welzel, 2005; Lupia & McCubbins, 1998 for some of the more notable contributions).
10


Taking into consideration the significant contributions of this body of work, the political behavior of policy actors specifically - as opposed to the general public - should be considered separately because policy actors operate under different conditions. For example, policy actors are ostensibly more experienced and may have more motivation for involvement; however, all policy actors are not able to participate in all activities or collaborate with whomever they please (Irvin & Stansbury, 2004a). A concerned citizen is not necessarily given the opportunity to provide testimony in Congress, regardless how much they may want to be involved. Public officials cannot participate in protests, regardless how much they may want to express their opinions. Nonprofits may simply not have the contacts nor the capacity required to develop policy networks with private industry counterparts, regardless how vital they may perceive that interaction to be in achieving their goals. There is differentiation among individuals as to which activities and policy networks are available, and this differentiation may be the result of contextual interaction with individual characteristics.
In summary, despite some theorizing and empirical work, there remains a lack of solid empirical evidence that explains what influences differentiation in individual political behavior among policy actors. This dissertation addresses that gap by contributing empirical research that explains how institutional context and individual characteristics interact to explain the variation of political behavior within the policy process.2 Political behavior within the policy process is crucial to our construction of political decision making. It is the amalgamation of the activities and network efforts of individuals involved in the policy process. Political behavior can provide input for decision makers; it can bring about learning and innovation - perhaps even policy
2 All types of participation in the policy process are defined as political behavior within this dissertation. Political behavior is not limited to the activities only, as the extant literature on participation would imply.
11


change - and is fundamental to democratic norms (Nohrstedt, 2013; O’Faircheallaigh, 2010). Through an enhanced understanding of political behavior, we better understand the resulting shape and intensity of political interactions, which are critical in a well-functioning democracy.
Conceptual Framework
This dissertation explores political behavior primarily through the lens of the Advocacy Coalition Framework (ACF). The Advocacy Coalition Framework (ACF) provides a useful lens for studying political behavior within policy subsystems, as the framework recognizes the role of individual behavior as a driving force within the policy process (Nohrstedt & Olofsson, 2016a). According to Nohrstedt, policy subsystems are artificial constructs that bound “analysis of participation in the policy process... subsystems are comprised of a multitude of actors involved in any given policy area” (2011), and these actors endeavor to impact policy through coordination of behavior in an intentional manner. By exploring policy subsystems as the basic unit of research, it is possible to design research that is generalizable to other contexts as well as other policy issues. The ACF explores three main areas: advocacy coalitions, policy change, and policy-oriented learning. The ACF endeavors to explain the policy process by understanding these three areas. This dissertation accepts the assumptions of the ACF regarding a modified version of methodological individualism and bounded rationality and will focus on exploring individual political behavior within coalitions.
Based on shared beliefs, policy actors organize into coalitions (Sabatier & Jenkins-Smith, 1993). Policy change can be brought about as policy actors within coalitions employ political behavior to impact the policy process (Blomquist, 2007; Jenkins-Smith, Nohrstedt, Weible, & Sabatier, 2014; Real-Dato, 2009; Weible, 2006). The ACF assumes that within coalitions, policy actors employ political behavior to impact outcomes. Activities and policy networks are the
12


means through which policy actors strive to influence the policy process (Heikkila, Pierce, et al., 2014). Political behavior is directed and constrained by the rules and norms of participation in the policy process, which are shaped by the political context of the subsystem (Boehmke, Gailmard, & Patty, 2005; Cross, 1999; Eisenberg & LoPucki, 1999; Guiraudon, 2000; Holyoke, Brown, & Henig, 2012; Howard, 2007; Jenkins-Smith et al., 2014; Karch, 2009; Ley, 2014; Pralle, 2003).
Numerous scholars recognize that the ACF is lacking in theory regarding how policy actors make political behavior decisions in varying contexts (Elgin, 2014; Jenkins-Smith et al., 2014; Nowlin, 2011; Weible, 2006; Weible & Sabatier, 2006), but the ACF does provide some direction about how context may be influential on differentiation in political behavior. Sabatier (1998) noted that there are attributes of the available settings in which policy actors may participate that can influence behavior. For example, a more prestigious forum forces professionals to participate. Jenkins-Smith explored this idea in early ACF theory, noting that “several key elements of the subsystem environment.. .appear to critically affect the manner in which analysts promote, refute, or adjust to potentially threatening analytical claims” (1990: 97). Jenkins-Smith recognized that the “policy environment” shapes the policy arena and its participants. The ACF assumes that within a contested subsystem, individuals form and maintain coalitions through which they work to influence the policy process (Sabatier & Weible, 2007). Empirical work drawing upon participation in extant ACF empirical research most often explores participation as a determinant in the formation of advocacy coalitions (c.f. Ingold & Varone, 2012; Nohrstedt, 2011) or as an explanation for policy change (c.f. Albright, 2011). There is also work on the determinants of “quality” participation (c.f. Webler, Tuler, & Krueger, 2001), but
13


this work focuses on the processes within participation and disregards the contextual factors that influence political behavior.
A clear need remains for further empirical and theoretical work in this area to consider the attributes of varying contexts and the impact on political behavior (Henry, Ingold, Nohrstedt, & Weible, 2014a). Against the backdrop of country variation, Henry et al. discussed the potential influence of differing contexts on policy outcomes (ibid.). There has been much discussion within the field regarding whether the ACF is applicable in different political systems (Henry et al., 2014a; Nohrstedt & Olofsson, 2016b, 2016a). These discussions recognize the role of subsystem context as influential in determining political behavior and ultimately policy outcomes, and the same argument can be extended to a sectoral level. As noted by Henry et al. “there are also significant variations between policy subsystems across time and space... situated within the same constitutional structure, policy subsystems display different opportunities and constraints for policy actors to access and influence the policy process” (2014: 302). There is not yet consensus on the types of contextual variables that may be most influential, and this dissertation aims to contribute to this discussion.
Different institutional contexts, expressed through sectoral differences, may also vary opportunities and constraints for political behavior. Differing contexts create different operating environments for policy actors, with potentially different strategies available to influence the policy process. For example, private foundations are expressly prohibited from lobbying by the Tax Reform Act of 1969. This is an institutional constraint imposed by sectoral affiliation within the nonprofit sector that shapes the opportunities private foundations have to influence the policy process, as recent evidence has shown to be increasingly occurring (Reckhow & Snyder, 2014). There has not been sufficient research on how different contexts may shape political behavior
14


within subsystems (Weible, 2006); supplemental theory is needed to move beyond what is currently recognized about the role of institutional context, and particularly its interaction with individual-level characteristics, in the ACF literature. To supplement the ACF and generate new theory, this dissertation will draw on institutionalism, behavioral economics, policy networks, advocacy literature, conflict expansion, venue shopping, and agenda setting.
Objective 1: Role of Individual Characteristics
To describe how individual characteristics shape environmental policy conflicts.
Understanding the policy process that underlies the development of environmental policy and the policy process, in general, requires understanding the individuals involved in environmental policymaking. According to the ACF, individuals are motivated to participate in policy activities and form coalitions to impact policy change owing to shared beliefs regarding, in this case, hydraulic fracturing. These individuals are boundedly rational and forced to call upon pre-existing belief systems to process their opinions and perceptions of a policy issue. Individuals often display shared beliefs and largely stable preferences regarding their policy positions and risk assessments. Importantly, individuals are motivated to form and maintain coalitions based on shared beliefs, and it is within these coalitions that policy actors endeavor to impact policy.
Objective 2: Activities
To identify patterns in the types of activities used by policy actors in the policy process and whether these patterns are influenced by institutional context and individual characteristics.
15


One fundamental tenet of the ACF is that coalition actors will use different activities to
influence policy (Weible & Sabatier, 2006). Activities are defined as the actions taken by policy actors to influence the policy process. Examples of activities are writing an editorial, sponsoring a bill, testifying in the legislature or filing a lawsuit. Activities are the opportunities for policy actors to engage in a subsystem in a variety of ways that are created and shaped by the rules and norms of participation. There are numerous ways in which activities of competing coalitions can differ. Within the ACF, activities are the byproduct of the resources of coalitions; the ACF assumes that coalition actors have specific resources that constrain their choice of strategy. According to Sabatier and Weible, coalition resources include: 1) legal authority, 2) public opinion, 3) information, 4) mobilization, 5) financial resources, and 6) leadership (Weible,
2006). This objective expands on this work recognizing context as influential in determining political behavior by exploring how activities are shaped by institutional context and individual characteristics. However, the research aims to identify many additional factors that may enable or constrain activities.
Since public policy began as a focused area of scholarship, researchers have been exploring the factors that influence participation in the policy process (Smith, 1984). Under the banner of conflict expansion, Real-Dato undertook a detailed theoretical synergy of four major policy process frameworks regarding their respective interpretations of the influence of context, particularly institutional context, as they relate to strategic participation in the policy process and ultimately affect policy change (2009). He asserted that context is a major determinant of participation decisions and called for empirical applications exploring how context impacts individual behavior within the policy process. Institutional factors, influenced by sectoral affiliation, may be central to the participation decisions of policy actors in a contentious policy
16


subsystem. Sectoral affiliation establishes the opportunities, constraints, and incentives of individuals within those institutional settings. The institutional context imposed on individuals in private versus public versus nonprofit sectors implies that the activities available to individuals in these differing sectors would vary.
Institutionalism has long called for explicit attempts to contextualize analysis and rational choice institutionalism searches for generalizable explanations of decision structures (Thelen, 1999), but few studies in the broader field of political behavior consider how context and individual characteristics might interact to influence political behavior. Sneiderman and Levendusky developed what they call an “institutional theory of public choice” (Sniderman & Levendusky, 2007). They combine the views of rational choice theory and the insights of political psychology to create a model of behavioral economics. The authors recognize that both external and internal factors are at play in an individual’s choice of political behavior; they emphasize that any full model of political behavior “requires two types of explanatory mechanisms - an internal one to account for choice between alternatives plus an external one to account for the alternatives on offer” (ibid.: 438). This objective adopts this same view: the political behavior alternatives on offer to an individual policy actor are determined in part by context and the choice among those alternatives is motivated by individual characteristics.
Objective 3: Policy Networks
To identify variation in policy networks and whether this variation can be attributed to differences in institutional context and individual characteristics.
Individuals are constrained by time, financial resources, capacity, and most importantly, institutional rules and norms that vary by sector. For example, there are regulations and norms
17


constraining the actions of nonprofit and public sector employees in regards to their efforts to impact the policy process. Owing to limited capacity for action, policy actors are forced to make strategic decisions and target particular individuals or organizations. The policy networks that arise within a policy subsystem may be driven by these constrained decisions. Policy networks can be viewed as the aggregation of many individual-level strategic choices. Policy actors could maximize their limited capacities by targeting those organizations and individuals whom they feel would have the highest potential impact on the policy process.
The third objective explores how institutional context and individual characteristics influence policy networks. Several scholars recognize the importance of policy networks (Agranoff & McGuire, 2001; Hill & Lynn, 2005; Lubell, Schneider, Scholz, & Mete, 2002; Milward & Provan, 2000; O’Toole & Meier, 2014; Provan & Milward, 2001). Much of the empirical policy network research focuses on collaborative networks under the guise of governance studies. This research often recognizes the role of constraints and competition within an advocacy space (see Lee, Lee, & Feiock, 2012), but does not advance a full network theory of the policy process as it fails to explain the characteristics of the components of a network - the individual (Dowding, 1995). This dissertation moves a step beyond by exploring not only with whom a policy actor chooses to connect but also the level of importance that policy actor assigns the interaction. By giving a measure of importance, the policy actors reveal the utility that he assigns that interaction, which can yield deeper insight about strategic network choices. With the individual as the unit of analysis, as opposed to the majority of studies done at the organizational level, this research can explore the institutional opportunities, constraints, and incentives in combination with individual characteristics that influence strategic network choices.
18


Consequently, we better understand the influence of institutional context and individual characteristics on a policy network.
There have been several studies of individual influence on strategic policy network formation (Hadden & Jasny, 2017; Johari, Mannor, & Tsitsiklis, 2006; Watts, 2001), but there are fewer studies that consider how the constrained capacity of policy actors influences their policy network. Leifeld and Schneider recognized the role of political opportunity structures and transaction cost analysis in individual decision making regrading who policy actors choose as contacts (Leifeld & Schneider, 2012). Using the case of information exchange, the authors determined that policy actors knowingly attempt to minimize transaction costs and maximize returns within their constrained capacity. This chapter builds upon this finding by capturing the strategic choices of policy actors in network connections, as influenced by institutional context and individual characteristics.
Hadden and Jasny recently explored how network embeddedness impacts policy actors’ political behavior (2017). The authors considered how an individual’s position in a policy network can influence their tactical choices. Looking at transnational nongovernment organizations (NGOs), results indicate that organizations are more likely to adopt a tactic if an organization with whom they are adjacent or directly connected has already adopted that tactic. Position in the network was found to be more influential than contextual equivalency, in that organizations were not likely to mimic similar organizations if those organizations were not directly connected, regardless how similar those organizations may be contextually. As the authors note, “social network theory draws attention to how interactions among actors shape their behavior” (2017: 1). In 2011, Ingold built upon previous studies recognizing the role of individual beliefs in network formation by studying if cooperative beliefs result in collaboration
19


(Ingold, 2011). The author argues that although beliefs were shown to be instrumental in predicting network cohesion, institutional settings may constrain or enable network ties and may in fact be a better predictor.
Lee et al. explored the structural features of an interorganizational network to answer questions surrounding the emergence of policy networks, the forces that shape network formation, and how networks evolve over time (2012). The authors found that organizations seek to build stable and sustainable relationships, which require more than simple exchange transactions. There appears to be some level of strategic awareness among organizations of the importance of participating in a dense network with credible ties. Theories of behavioral economics adopt tools of economic analysis to explore, for example, political issues. Behavioral economics assumes that individuals act strategically and in their self-interest to maximize their utility (Downs, 1957; B. D. Jones, 2001; Shepsle, 2008). These assumptions are not limited to economic choices; this applies political behavior in choice of network targets as well. Theoretical economic models of individual strategic choice in network formation have been developed (Bloch & Jackson, 2006; Galeotti, Goyal, & Kamphorst, 2006; Goyal & Vega-Redondo, 2005; Jackson & Watts, 2002; Jackson & Wolinsky, 1996) but empirical applications are lacking.
To better understand micro-level political behavior within the policy process, this objective focuses on another aspect of political behavior: policy networks. The unit of observation is again the individual policy actor, but this objective will explore a policy network as individuals strategically choose which organizations to interact with in order to achieve their policy goals. Many studies in empirical ACF work use networks to identify coalitions (Fischer, Ingold, Sciarini, & Varone, 2012; Lienert, Schnetzer, & Ingold, 2013), often recognizing the role of shared beliefs as the cohesive mechanism for network formation (Ingold, 2011; Matti &
20


Sandstrom, 2013; Weible, 2005; Zafonte & Sabatier, 1998). As Ingold notes, coalitions are often identified based on “the assumption that common beliefs are reflected in relations among actors involved in policy processes” (2011: 435). The research proposed here does not dispute that shared beliefs can encourage the formation of policy coalitions as networks but rather aims to bring more theoretical clarity to that assumption by testing whether tie formation might also be motivated by context. There is substantial empirical evidence that belief hornophily is a driver of network formation (Jenkins-Smith et al., 2014); this research seeks to explore other contextual variables, such as institutional conditions, that may influence network choices as well.
This objective is built upon previous network studies in the ACF to expand knowledge regarding network formation, exploring how an individual policy actor’s institutional context and individual characteristics may influence network structure. The goal of this objective is to identify and explore political behavior as networks reflecting who an individual policy actor targets to achieve his or her policy goals. These are related to but may also be distinct from coalitions. This study identifies coalitions as a basis for the network analysis, then explores further differentiation into policy networks based on sectoral affiliation. This parsing alone yields some explanation regarding how, for example, the policy network of nonprofit-affiliated individuals is structurally different than the policy network of private sector individuals. The policy networks in this study are not intended to represent or define coalitions3, but rather will be differentiated by institutional conditions and individual characteristics.4 In this study, policy
3 There are several reasons why this particular network analysis does not intend to define or represent coalitions. First, the network is unidirectional and not symmetrical, in that the ties are not necessarily reciprocal. Second, the network is also bipartite, or two-mode, composed of two different types of vertices, in this case individual-organization. Finally and relatedly, the network is incomplete, as is common in social network analysis.
4 It should be acknowledged that some network studies have defined and identified coalitions using the type of network described proposed here. There is a sizeable literature within the ACF that debates how coalitions should be defined: by beliefs, or by network connections, or by shared venues, among others. Whether network connections,
21


actors holding opposing positions may find themselves in the same policy network when conditioned on institutional context such as sector instead of on stance. The ACF is a framework that enables multiple theoretical arguments (Jenkins-Smith et al., 2014), and while the goals of this research are somewhat novel within the framework’s empirical work, it still conforms to ACF assumptions. There is overlap in this question compared to what is traditionally asked in ACF coalition studies.
Research Design
This dissertation is multi-method case approach that aims to inform about a larger group of cases: high-conflict policy issues in general (Gerring, 2007a). The goal is to inform the larger from the smaller, generalizing the explanations that result from this research to the broader population of high-conflict policy issues. While the ultimate aim of the research is explanatory, which often precludes “thick” case study description, this research could also be characterized as a thick by Blatter and Blume (2008), as opposed to Gerring. According to Gerring, “researchers invariably face a choice about knowing more about less, or less about more” (2007: 49) but Blatter and Blume argue instead that thickness should not immediately dictate a tradeoff in generalizability.5 Thickness should instead be determined by a multiplicity of observations within the case as well as the researcher’s attention to theoretical explanation of the relationships discovered. It is in these cases that applicability of our findings to broader research paradigms can be maximized, and within which this particular case study aims to contribute.
particularly unidirectional network connections, are sufficiently cohesive in coalition formation, remains an open empirical question (see Henry, Ingold, Nohrstedt, & Weible, 2014) but not one fortius dissertation.
5 Blatter and Blume (2008) explore the covariance form of case studies, in which variables-centered testing dominates the analysis. Objectives 2 and 3 of this dissertation conform to this form. As the authors explain, “covariance is aiming to draw generalizing conclusions from cases to a wider population” (ibid.: 317).
22


Case Selection
With its a multi-method case structure and aim to uncover patterned influences of context, the dissertation is novel in its approach to first use data from the national level without regard for contextual variation. This sets up the initial understanding of the role of individual characteristics. The study then moves into the specific case of Colorado, which is a typical case and by definition is representative (ibid.). Identifying a case as representative requires some knowledge about the characteristics of the broader population. A representative case of high-conflict policy issues should demonstrate continuity in the salience of the issue, maturity of the subsystem, and variation in political ideology and theoretically-driven expectations on factors that might dampen or bolster conflict.
Hydraulic fracturing has been in use in Colorado for decades, and while oil and gas revenues are sizeable enough to make the issue highly salient to the general population, the state is not solely dependent on revenues from production. In contrast, a state such as Texas would be a poor choice as a typical case as it is much more dependent, relatively, on oil and gas production. Texas would be better identified as an extreme case. There are areas within Colorado that are more dependent on oil and gas revenues than other areas but this is important to the expectation of issue salience. It is reasonable to expect that political ideology may contribute to levels of conflict on policy issues, and hence variation in political ideology is important for a typical case. Political leanings in Colorado have been fairly evenly split for the last century, with nearly the same number of Democrats and Republicans elected as Governor, for example (Bump, 2015). As with much of the rest of the country, there is an ideological divide between urban and rural areas, with urban areas and some mountainous regions pulling the state’s overall ideology
23


slightly left, but the state has been a swing state in the last two presidential elections and is expected to continue to be relatively open ideologically.
This research uses an exploratory and explanatory approach to understanding the case of hydraulic fracturing in Colorado. The aim of the research is to uncover and explain the political behavior expressed by policy actors in different contexts. A non-experimental case study, such as this, is designed to explore the limits of existing theory and possibly build new theory (Van de Ven, 2007). Van de Ven notes that non-experimental case studies may be useful in construct validity, drawing upon theoretical inference to develop better constructs within our theories. These types of studies are “less interested in the accuracy of prediction per se than what a relationship reveals about the meaning of the concepts being measured” (Singleton & Straits, 2010: 141). This case meets the preconditions of the covariance approach, as first detailed by Gerring (2007a) and further explicated by Blatter and Blume (2008). The approach demands control of exogenous variables through case and indicator selection and theoretically-driven hypotheses regarding the causal direction of relationships. Case studies applying the covariation approach observe, over time and/or space, variance in the independent and dependent variables of interest. The conclusion is therefore that X has a causal effect on Y. Covariation case studies are focused on theory, not on explicit aims of explanation of the case itself (Blatter & Blume, 2008). The covariance case study seeks not to achieve consistency within empirical results but rather within more abstract concepts. This is crucial to claims of generalizability and theorytesting. Therefore, the purposive choice of the case of hydraulic fracturing in Colorado aims not to explain the exploration of unconventional oil and gas but rather to use the insights gained through this case setting to explain political behavior within high-conflict policy issues in general.
24


Methods of Data Collection
The analysis draws upon a set of surveys and interviews completed at the national level in 2016 and in Colorado in 2015 and 2017. The in-depth interviews associated with the surveys served three separate purposes. First, the interviews informed the development of the survey. Second, the interviews helped identify additional policy actors and activities not picked up in the mapping process as a form of modified snowball sampling. Snowball sampling was developed as a way to uncover the structure of social networks for which the population is not known a priori (Douglas D Heckathorn, 2011). Modified snowball sampling attempts to correct for the issue of hidden populations inherent in any social network. Previous experience within the policy issue generated an initial set of interviewees, after which modified snowball sampling was used to provide names of any other individuals that key stakeholders felt were important to interview as well. Finally, the interviews provide richer context for the quantitative data gathered through the survey. A mixed methods approach in case study research design can be more impactful in this simultaneously exploratory and explanatory study (Riccucci, 2010; Van de Ven, 2007).
Sampling: Policy Actor Identification
The first step in any ACF application is to identify the policy actors involved in the subsystem. One of the basic assumptions of traditional ACF theory is that it adopts a viewpoint of a decade or more. Weible and Sabatier have noted that the ACF can consequently be difficult to apply as it is time consuming and challenging to gather details about the actors involved over such a long timeframe (2006). ACF scholars have encouraged researchers to develop creative, more accessible, and possibly quicker methods to map policy subsystems (Jenkins-Smith et al., 2014; Weible & Sabatier, 2006).
25


This research first mapped the policy subsystems to uncover the policy actors involved in the subsystem through online identification and reporting of involvement in the policy issue.
This method utilizes advancements in technology and worldwide communication since the inception of the ACF. However, there are limitations to the method, although many of the limitations apply to traditional mapping methods within the ACF literature as well. First, it can be difficult to identify hidden or silent policy actors in the subsystem without expert knowledge of the subject area. Second, this method of subsystem mapping relies exclusively on publicly available reports and accounts of actor activity. There are other ways in which policy actors participate in the subsystem that are not publicly available. This limitation will be addressed in later steps of the data collection process. Third, the researcher must have at least a basic understanding of the context and policy processes in the subsystem in order to uncover the activities of the actors.
Social science research has begun to recognize the usefulness of the internet as a tool (Singleton & Straits, 2010). The subsystem mapping and policy actor identification started with an online search of the policy issue to include legislative proceedings, court proceedings, media and any other venues in which policy activity is possible. The search terms were “Colorado/United States AND [hydraulic fracturing OR fracking OR unconventional oil OR unconventional gas OR shale oil OR shale gas]”. Those individuals who were reported as active within the policy issue were identified as potential policy actors. Media - including but not limited to blogs, reports, press releases, articles, and conference presentations - was also searched to identify active individuals. Any publicly available document that records evidence of participation was used. Finally, individuals and organizations now identified were cross searched with the original search terms to explore other activity that was not initially found.
26


The sample for the survey was purposive. The survey sample is not intended to be a reflection of public opinion; it is intended to be a survey of the perceptions and attitudes of individuals who are already engaged in the policy debate. Survey questions were developed after in-depth interviews with key policy actors in each state. The survey was reviewed in its entirety by an advisory board of influential and experienced individuals who are highly engaged in the topic of hydraulic fracturing. See Table 1 for a summary of the data collection results of the two waves of the surveys. The unit of analysis is the individual. Some predictors are measured at the organizational level, but these are considered to be influences upon the individual’s political behavior decisions and do not therefore necessitate a need for clustering.
Table 1: Data Sources
Chapter Geographic Scope Source (n) Date Gathered Level of Observation Contents Analysis
Chapter 1 National Survey (133) Interviews (ID Spring 2016 Individual Activities Network Context Beliefs Stance Threats Controls - Statistical description - Basic correlations
Chapters 2 and 3 Colorado Survey (213) Interviews (9) Spring 2015 Individual Activities Network Context Beliefs Stance Threats Controls - Factor analysis - Multinomial logistic regression - Social network analysis - ERGM - Qualitative triangulation
Chapters 2 and 3 Colorado Survey (186) Interviews (29) Spring 2017 Individual Activities Network Context Beliefs Stance Threats Controls - Factor analysis - Multinomial logistic regression - Social network analysis -ERGM - Qualitative triangulation
27


Key Variable Operationalization
This research uses Gerring’s minimal definition strategy. This strategy “seeks to identify the bare essentials of a concept, sufficient to differentiate it extensionally without excluding any of the phenomena generally understood as part of the extension” (Gerring, 2012: 135). Because the aim of this research is to explore all possible combinations of political behavior used by policy actors to influence the policy process, a minimal definition of variables is preferred. In addition, many of these concepts are nebulous and adopting a parsimonious definition with widespread understanding can be an effective strategy (Gerring, 2012). Appendix A details constructs and operationalization for all key variables.
Activities will be operationalized as an action that an individual takes to participate in the policy process with the intent to achieve their personal or professional goals in related to oil and gas development that uses hydraulic fracturing (refer to question 17 in Appendix B: Full Survey). Each respondent was given a choice of eight different activities and asked to rank the effectiveness of those activities on a Likert-type scale from Not Engaged to Very Effective. The activities were: Brokering agreements between parties, Countering arguments made by people you disagree with, Mobilizing the public, Collaborating with people you disagree with, Coordinating political activities with allies, Providing information to government officials, Providing information to the news media, and Sharing your opinion with government officials. Previous versions of the survey asked respondents to identify how frequently they participate in a similar battery of activity options. The question wording was updated to capture a respondent’s perception of the effectiveness of different activities, instead of limiting the information gathered to simply a measure of frequency. The spectrum of activities was vetted through interviews, reviewed by an advisory council of issue experts, and tested in previous versions of the survey to
28


ensure that it represented a reasonable majority of possible ways in which a policy actor might engage in the policy process.
Policy networks are the networks of organizations and individuals that are accessed by policy actors to achieve their personal or professional goals related to oil and gas development that uses hydraulic fracturing. Respondents were given a battery of fifteen organizations -Federal government, State government, County government, City government, Oil and gas industry, Oil and gas industry, Environmental or conservation groups, Real estate developers or home builders, Agricultural organization or farmers, Organized citizen groups, Churches or other religious organizations, Universities or colleges, Consulting firms or think tanks, Informal personal networks, and News media - and asked to rank importance of interaction on a Likert-type scale from not at all important to very important, with an option for not applicable (refer to question 9 in Appendix B: Full Survey). As with the activities question, this battery of choices was tested in earlier versions of the survey and reviewed by an advisory council. Finally, political behavior is operationalized as the amalgamation of activities and policy networks.
Analysis
The result of the data collection is a collection of rich qualitative and quantitative data that can be used to expand theory on the nature of political behavior in different contexts. The study illuminates the types of behaviors that could emerge and provide information about how individuals choose one activity or network collaborator over another. The simultaneously descriptive and explanatory nature of the research question lends itself to rich description of the results and may also lead to the creation of new typologies of political behavior. Nuances of the interaction between the organizational context and individual characteristics may be used to illuminate the nature of political behavior as well as the context variables themselves. Theory
29


expansion can be achieved by using the analysis to revisit the original propositions and develop hypotheses of a more explanatory quality related to what drives political behavior decisions.
In the first objective and Chapter 1, statistical description and basic correlation is used to describe the role of individual characteristics in shaping environmental policy debates. The research explores the beliefs and policy preferences of individuals involved at the national level in the policies shaping the extraction of unconventional oil and gas in the US; it focuses on national-level policy actors as these individuals - in comparison to individuals active at only the state or local level - should be less susceptible to local-level variance.
For the second objective and Chapter 2, factor analysis is used to determine if there are patterns among activity choice by policy actors. Determinants of those activity patterns, hypothesized to be stance on the issue, perceptions of viability, and perceptions of threat, are subsequently tested through multinomial logistic regression. Gender, education, professional experience, stance, and political ideology serve as controls.
For the final objective and Chapter 3, social network analysis is used to first describe the policy networks that have developed around hydraulic fracturing in Colorado. The whole network is then differentiated by sectoral affiliation (public v. private v. nonprofit) of the respondent and network structure is regressed against expectations regarding strategic network choices using inferential network analysis through an exponential random graph models (ERGM). The ERGM results of each sectoral network are compared to explain variation in network structure across sectors.
30


Limitations
Case study research inherently faces tradeoff of external validity versus representativeness. Generalizability can be limited owing to the small number of cases under consideration (Gerring, 2007b). This study attempts to correct for this limitation by designing research at the subsystem level and observing within the subsystem, which may lend itself to application in other contexts. In addition, the purposive choice of a representative case enhances the potential for generalizability. The findings may lend themselves to empirical generalizability to other high conflict environmental policy issues, but there is potential that theoretical generalizability could be limited to other cases in which institutional context is determined by sector. There is a risk of collinearity among the key political behaviors. However, since this study does not aim for precise causal explanation of the empirical setting but rather general theoretical observations, collinearity can be managed (Gerring, 2012).
One limitation of many ACF applications is the inability to locate hidden populations. There are two aspects to hidden populations. First, the study might be limited in its ability to locate or access hidden populations of policy actors who do not appear in the sampling frame. As an attempt to address this issue, identification of potential hidden populations was emphasized in the modified snowball sampling in the interviews. Relatedly, there is potential for selection bias in the sample. As the sample is intended to identify those individuals who are active in the policy process, there may be an overrepresentation of frequency of participation or a structural tendency to survey individuals whose cognitive characteristics orient them towards participation.
The second aspect of hidden populations, however, may be more of a contribution than a limitation. Recent work has explored the “dark side of public policy”, arguing that the hidden agendas of policy actors may in fact “actually serve the business of governing” (McConnell,
31


2017: 1). These policy actors may be visible but their actions may not, which is part of the population identified in this study. This research project aims to explain why policy actors chose the strategies they do. There is a story behind why some policy actors chose to remain hidden, which is a strategy itself and may be the result of specific political, subsystem or individual resource context.
Contributions
While past research has explored individual components of political behavior, this dissertation offers a more holistic picture of how an individual behaves within the policy process. Research exploring how political behaviors vary in different contexts and across time would bolster the explanatory capacity of the ACF. Expanding theory on political behavior contributes to the broader ACF framework and the overall policy process literature. With more knowledge and empirical evidence regarding the causal processes underlying political behavior, it is possible to refine theory to address contextual variations. The research contributes to ACF empirical work exploring and explaining how political behavior may vary based on the circumstances under which a policy actor finds himself. Research exploring how political behaviors vary in different contexts would bolster the explanatory capacity of policy process theories. This research also has implications for understanding policy change. In theory, policy actors choose “winning” strategies within sympathetic venues in which those strategies are most likely to prevail (Holyoke et al., 2012; Schattschneider, 1960). To understand influence within the policy process, it is important to understand the behaviors individuals access to create that change and whether there are systematic barriers that shape political behavior, which in turn impacts the quality of politics. Owing to institutional conditions, particular groups or individuals may be disadvantaged within the policy process. This dissertation can contribute to
32


understanding whether disparities in political behavior are the result of structural inequalities deriving from institutional context or rather due to individual characteristics. The former is less encouraging for a well-functioning democracy.
Methodologically, the research proposed will utilize a variety of data analysis tools to explain the influence of context on political behavior. Many of these statistical approaches, such as inferential network analysis, are not currently widely used within the ACF. There are very few applications within policy process studies that use longitudinal data to employ inferential network analysis; this dissertation is an infrequent example of longitudinal empirical work on political behavior.
Through the dissertation’s findings, we better understand not only what a policy actor does to impact the policy process, but with whom the policy actor interacts and the importance of interaction to better understand the governance landscape of this high-contested policy issue. The first objective describes the role of individual characteristics in shaping highly-contentious policy issues and sets the stage for the following two objectives. Objective 2 produces knowledge regarding patterns of activity choice, as differentiated and determined by institutional conditions and individual characteristics. Objective 3 draws upon a novel approach to network studies and considers how institutional constraints and individuals characteristics strategically shape policy networks and influence the efficiency of the resulting networks. The dissertation as a whole develops nuances to the interaction between institutional- and individual-level characteristics, and ultimately, informs theory and practice regarding the consequences of contentious policy exchanges. In terms of practice, this research could be used by public sector individuals, nonprofit organizations, and private enterprise to better understand how their specific context may enable more effective participation in the policy process.
33


CHAPTER II
PERCEPTIONS OF CONTENTIOUSNESS: HOW INDIVIDUAL TRAITS SHAPE ENVIRONMENTAL POLICY CONFLICTS6
The extraction of unconventional oil and gas using hydraulic fracturing is a complex environmental issue that has drawn contention from all branches of government in the United States (US) - Congress, the White House, and numerous federal agencies - as well as state government and local municipalities. The Obama administration fought to regulate hydraulic fracturing on federal lands, an important step for national regulation. However, while many applauded the Obama administration’s attempts to regulate, many others felt the efforts were shallow and did not go far enough (Harder & Gilbert, 2015). Regardless, Obama’s hydraulic fracturing regulations were quickly reversed by the Trump administration (Tabuchi & Lipton, 2017). The Environmental Protection Agency (EPA) spent more than six years on a landmark study exploring the water-related impacts of hydraulic fracturing. When the agency’s report was released in 2015, it was met with contempt from groups both opposed to and in support of hydraulic fracturing. The former felt the EPA’s recommendations were not well founded and exposed citizens to potential danger, while the latter were disappointed in the lack of an enforceable directive eliminating future questions related to the environmental impacts of oil and gas extraction (DiChristopher, 2016). By December 2016, the EPA reversed course somewhat and highlighted the sometimes tenable but nonetheless real connection between oil and gas operations using hydraulic fracturing and contamination of drinking water (EPA 2016). While oil and gas production utilizing hydraulic fracturing may have tapered off from its apex in 2015, the
6 Portions of this chapter were previously published in Environmental Policy and the Pursuit of Sustainability (2018) London, United Kingdom: Routledge, and are included with the permission of the copyright holder.
34


debate surrounding the extraction of unconventional oil and gas remains critical for environmental and energy policy (Heikkila, Weible, & Olofsson, 2017).
The extraction of unconventional oil and gas using hydraulic fracturing hits on many environmental issues including water supply and quality, air quality, emissions, public nuisance, energy security, economic impacts, energy costs, and public health matters. The issues involved are complex and require involvement from all levels of government, and consequently, environmental policies in pursuit of sustainability are difficult to form. Understanding the policy process that underlies the development of environmental policy and the policy process, in general, requires understanding the individuals involved in environmental policymaking. The objective of this research is to understand the beliefs and policy preferences of individuals involved at the national level in the policies shaping the extraction of unconventional oil and gas in the US, including individuals involved in policymaking, citizens impacted by hydraulic fracturing, individuals who work in hydraulic fracturing professionally, government regulators, environmental groups, interested media, and more. The research focuses on national-level policy actors as these individuals - in comparison to individuals active at only the state or local level -should be less susceptible to local-level variance. The purpose of this research is to understand how the individual-level beliefs and policy preferences of actors in environmental decisionmaking influence their views while participating in the policy process. By exploring the beliefs and preferences of individuals involved in a high-conflict policy issue, and how those belief structures shape perceptions, we can better understand the development of environmental policy conflicts. After a discussion of the case setting, the chapter first examines the characteristics of national-level policy actors involved in making policies regarding hydraulic fracturing and then
35


explores how differences in characteristics among policy actors are associated with varying perceptions of issue contentiousness.
Case Setting: Hydraulic Fracturing
In hydraulic fracturing, fluids and other chemicals are injected into wells at high volumes and pressure to fracture the rock and release previously “trapped” oil and gas for collection (Rahm, 2011). The technique of hydraulic fracturing is not new; its emergence dates as early as the 1940s (Hubbert & Willis, 1957; Phillips, 1972). The use of hydraulic fracturing increased dramatically around 2005 as advancements in horizontal drilling technologies opened up ways for the oil and gas industry to access shale oil and gas reserves that were previously not economically viable (Heikkila, Pierce, et al., 2014). As the usage of the technique rapidly expanded, so too did the debate. Moratoriums and bans have been discussed in several US states, from New York and Pennsylvania to Oregon. Bans have been passed in three states to date -New York, Vermont, and Maryland, as recently as March 2017 - with the ban in New York in 2014 garnering significant national attention. Moratoriums and local-level bans have been contested in numerous highly visible cases in several state supreme courts.
The hydraulic fracturing policy issue is not only important for environmental and energy policy conflicts, but it is also critical to understanding the behavior of individuals involved in high-conflict policy issues. Hydraulic fracturing is marked by complex policy issues, interrelated and often inseparable motivations of policy actors, relevancy at multiple levels of government and regulatory agencies, and broad importance to the general public as an issue that is directly related to the economy and the daily lives of many citizens. The results of this study can help in drawing more general conclusions about high-conflict environmental policy issues. It is critical for policymakers and citizens alike to understand high-conflict environmental policy issues
36


because there are implications for the quality of politics that emerge in these situations and related democratic and environmental justice concerns. In high-conflict policy issues, there may be a tendency towards polarization and stalemate, which is not productive for effective or quality policy. In terms of democratic concerns, we should study environmental policy conflicts because we learn more about who is participating and how, if there are barriers to participation, and whether there is an unjust distribution of benefits or unequal burden of risk. Issues related to hydraulic fracturing are fraught with concerns regarding environmental justice. Very few individuals are directly impacted by hydraulic fracturing operations, yet many, many more individuals experience the benefits and have the power and/or access to impact policy. While this research does not endeavor to address all questions related to hydraulic fracturing policy, it does contribute to understanding how individual beliefs and policy perceptions shape an environmental policy conflict.
The theoretical basis for the research is the Advocacy Coalition Framework (ACF). The ACF is a useful lens through which to view the development of high-conflict policy issues. According to the ACF, individuals are motivated to participate in policy activities and form coalitions to impact policy change owing to shared beliefs regarding, in this case, hydraulic fracturing. These individuals are boundedly rational and forced to call upon pre-existing belief systems to process their opinions and perceptions of a policy issue. We can expect, therefore, that individuals will display shared beliefs and largely stable preferences regarding their policy positions and risk assessments. Importantly, individuals are motivated to form and maintain coalitions based on shared beliefs, and it is within these coalitions that policy actors endeavor to impact policy. This can become problematic for effective policy making if coalitions become silos and perceptions of high-conflict issues become both polarized and stagnant. This research
37


aims to interrogate the structures and perceptions of policy coalitions surrounding hydraulic fracturing in the US to explore how these coalitions are impacting the policy process.
Data Collection
Data collection started with a series of eleven in-depth interviews during autumn 2015 with policy experts in unconventional oil and gas extraction. The policy experts were from a range of organizations, including professional industry association representatives, federal agencies, nonprofit environmental groups, advocacy groups, and government representatives. There was no geographic target for these interviews, but all interviewees were active in policy at the national level, as opposed to the state or local level, in matters concerning the extraction of oil and gas using hydraulic fracturing. The in-depth interviews were used to tailor the survey to reflect the shape of the issue at the national level, and the survey was then vetted with an advisory committee of policy experts and updated based on their recommendations. These interviews also contributed to development of the sampling frame.
A survey of policy actors involved at the national level on issues related to the extraction of unconventional oil and gas using hydraulic fracturing was administered in the spring of 2016 (see Appendix B for the full survey). A total of 468 policy actors were identified as actively involved in or knowledgeable about unconventional oil and gas development in the US. Individuals were included without regard for organizational affiliation or profession. Rules for inclusion required that the individual had to demonstrate activity within the policy issue since 2008 and had to be currently active at the national level (as opposed to the state or local level). Similar sampling procedures have been used in past research at the state and local levels (Heikkila et al., 2017; Weible, Olofsson, Costie, Katz, & Heikkila, 2016). The issue of hidden populations is a limitation of any sampling procedure in which the population is not known
38


apriori, as is common in social science research (D. D. Heckathorn, 1997). Because there is no comprehensive register of policy actors, approximation methods must be used to identify the population of individuals that may be active within the issue and to review the composition of the final sample to explore possibilities of systemic errors in the creation of the survey sample. Individuals were first identified through the in-depth interviews and then used as seeds for online discovery of activity (Nohrstedt & Olofsson, 2016b). For online discovery of activity, the internet is searched for any and all forms of activity as potential for inclusion: testifying at a public hearing; comments on a public forum; employment or involvement with private, public, or nonprofit organizations active within the policy space; social media; traditional media, and many more. For example, if an individual with a federal agency mentions during the interview that she worked with a research group at a university, the website of the research group is found and then reviewed to potentially include any and all individuals within that group that are active within the policy issue of hydraulic fracturing at the national level. If perhaps the interviewee was connected to the university research group through a co-worker in another department, then that co-worker is identified for inclusion in the survey sample and the department’s employees are all reviewed for potential inclusion as well.
The survey was administered via email using the Qualtrics platform. Of the 468 individuals who were administered the survey, 133 responded for a response rate of 28%, which is considered acceptable for online social science survey research (Nulty, 2008; Topp & Pawloski, 2002). The design of the survey was theoretically-based on the ACF and drew on similar surveys administered in the same issue area (see summaries in Heikkila et al. 2013; Heikkila & Weible 2015). Survey research is a common tool within the framework’s empirical
39


tradition. The survey was developed under the framework to uncover the beliefs, positions, and perceptions of policy actors involved in the issues surrounding hydraulic fracturing.
Results
There does not appear to be systematic bias in the organizational affiliation or position of respondents, limiting the potential for sampling bias. In comparison to the full sample, there was an equal spread of organizational affiliations among respondents. In terms of stance, it is nearly evenly distributed; respondents with a stance against hydraulic fracturing report that they would like to see hydraulic fracturing either stopped (16.8%) or limited (34.5%) in comparison with respondents with a stance in support of hydraulic fracturing that prefer production continue at the current rate (25.7%), expand moderately (14.6%), or expand extensively (8.4%). The average survey respondent holds at least a Bachelor’s degree, tends toward moderate to slightly liberal political views, and describes his priority level regarding dealing with issues related to hydraulic fracturing in either a professional or personal capacity as “high”.
The average respondent has relatively high levels of experience in reading about issues related to hydraulic fracturing and slightly less experience researching and analyzing topics involved in hydraulic fracturing, with comparatively low levels of experience in planning, working, or managing oil and gas operations or living within visible proximity of oil and gas operations. Combined with the relatively high level of education within the sample, the variation in types of experience with issues surrounding hydraulic fracturing may indicate that this sample is more professionalized in their experience rather than experiencing the issue personally.
Finally, the average respondent holds a position that oil and gas development that uses hydraulic fracturing should be limited somewhat or continued at the current rate, in comparison to other
40


respondents who take more extreme positions that hydraulic fracturing should either be stopped altogether or expanded moderately or extensively.
There are some sizeable variations within these averages, however. Regarding education, the sample mode is for a Masters or professional degree but that only represents one-third of the sample. Nearly as many individuals hold either a Bachelor’s Degree or a Ph.D. or J.D. While nearly half of all respondents reported that dealing with hydraulic fracturing issues is of high importance to their professional or personal goals, nearly one-third reported that it is only of low to moderate importance. As noted earlier, the vast majority of the sample holds either liberal or moderate political views, with very few respondents reporting conservative or extremely liberal views. The modal category of organizational affiliation of the respondent was Environmental or Conservation Groups, representing around 38% of the sample. This was followed by Consulting Firms or Think Tanks (15%), Federal Government (11%), Oil and Gas Industry (11%), Oil and Gas Professional Associations (7%), University Research (7%), Organized Citizen Groups (4%), State Government (2%), and Other (5%).
Respondent Belief Structures
We can now begin to explore the belief structures of these policy actors and associate those with our variable of interest: individual perceptions of the conflict, measured as a respondent’s opinion regarding the level of contentiousness of the policy issue of hydraulic fracturing. Respondents were asked to rate the level of political contention regarding oil and gas development using hydraulic fracturing, in comparison with other political issues in the United States. The Likert-type scale ran from 1 to 5 with the following response options: Far less contentious (1), Less contentious (2), Just as contentious (3), More contentious (4), Far more contentious (5). Table 4.2 summarizes the results.
41


Table 2. Perception of Level of Issue Contentiousness
Frequency Percent Cumulative Percent
Far less contentious 3 2.80% 2.80%
Less contentious 16 14.95% 17.76%
Just as contentious 42 39.25% 57.01%
More contentious 36 33.64% 90.65%
Far more contentious 10 9.35% 100%
TOTAL 107 100.00% 100.00%
The most common response was “Just as contentious,” representing nearly 40% of the sample. The mean response for this question was 3.32 with a standard deviation of 0.94, indicating that generally, respondents find hydraulic fracturing to be at least as contentious if not more contentious than other political issue in the United States. Only about 18% of respondents felt it was less contentious or far less contentious. The standard deviation is relatively high in comparison to the mean, which indicates that there is fairly wide variation in the answers. However, that variation is concentrated within the group of respondents who feel the issue is at least as contentious or worse, which makes the distribution of responses largely irrelevant for our purposes. For ease of interpretation, responses on perception of level of issue contentiousness were collapsed into two categories: Less Contentious (n=61) and More Contentious (n=46). Figure 4.1 presents the individual traits discussed in the previous section, differentiated by perception of contentiousness.
42


35
30
25
20
15
10
5
0
25
20
15
10
5
0
Stance Political Position
30
Less Contentious More Contentious Less Contentious More Contentious
â–  Pro Stance I Anti Stance â–  Liberal I Moderate â– Conservative
Education Experience
30
Less Contentious More Contentious Less Contentious More Contentious
â–  B.A. i Master's Degree â–  Less Experience â–  Moderate Experience
â–  Law Degree â–  Ph.D. or M.D. â–  Eligh Experience
Figure 1: Individual Traits by Contentiousness
It is encouraging that there are not strong patterns within the sample regarding stance, political position, and education in association with a respondent’s perception of issue contentiousness. Strong patterns in these basic individual traits such as education and stance might indicate sample bias or that the particular policy issue of hydraulic fracturing is defined by its complexity or along partisan lines. There appear to be some differences regarding experience associated with an individual’s perception of the level of contentiousness of hydraulic fracturing, but it is important to note that none of the differences between and among groups were found to be statistically significant through two-way measures of association (Pearson’s Chi2 test).
43


In absolute terms, however, one of the more noticeable patterns is in professional and personal experience, which has also been collapsed into three categories for ease of interpretation. Individuals with less experience have a tendency to perceive hydraulic fracturing as less contentious, in comparison with individuals with moderate or high experience, who have a tendency towards a perception that the issue is more contentious. Caution is warranted in drawing generalizations regarding the association between experience and contentiousness, as this measure is rather blunt. It collapses several different types of experience, such as researching or analyzing the impacts of hydraulic fracturing, into one category with other types of experience, such as owning mineral rights or living in visual proximity of oil or gas operations.
To check for construct validity of the collapsed measure, all individual types of experience were tested individually against contentiousness in two-ways measures of association using Pearson’s Chi2 test and two were found to be statistically significant: “Regulating or governing oil and gas development” (p<0.10) and “Participating in political activities to influence government decisions about oil and gas development” (p<0.05). The omega-squared effect size for regulating experience indicates that this type of experience accounts for nearly 4% of the variance in perception of issue contentiousness, which appears relatively small but is still statistically significant. Even more pronounced is the omega-squared effect size for participating in political activities, which indicates that this type of experience accounts for more than 10% of the variance in perception of issue contentiousness. It appears that experience with particular types of activities does have an impact on an individual’s perception of issue contentiousness.
We can now begin to explore the belief structures of individual respondents to understand how cognitive characteristics, such as perceived threat, may be associated with perceptions of issue contentiousness. Recent work on policy conflicts has shown that individual
44


traits like education and stance are not as consistently influential upon perceptions of contentiousness as are cognitive characteristics such as threat perception or dissenting relationships (Heikkila & Weible, 2017; Weible & Heikkila, 2017). Figure 4.2 illustrates the associations between four cognitive characteristics and perception of level of issue contentiousness.
Personal Threat**
Less Contentious More Contentious
â–  Very Little â–  Quite A Bit
Relationships*
Less Contentious More Contentious â–  Not Collegial i Somewhat Collegial
a Very Collegial â–  Completely Collegial
Theaten United States*
i
Less Contentious More Contentious
â–  Very Little Quite A Bit
Less Contentious More Contentious â–  Low Priority â–  High Priority
Note: Statistical Significance: ***p<0.01, **p<0.05, *p<0.10
Figure 2: Cognitive Characteristics by Contentiousness
All of the associations in Figure 4.2 were found to be statistically significant, indicating that there are non-random relationships between these cognitive characteristics and perception of
45


issue contentiousness. Respondents were asked to gauge how the actions of those you disagree with on oil and gas development that uses hydraulic fracturing threatens both themselves and the Unites States. Response options were on a Likert-type five-point scale from “not at all” to “a great deal.” The mean for personal threat was 2.57 and much higher for threatening the United States at 3.40; however, when associated with issue contentiousness, differentiation is more pronounced for personal threat. Policy actors appear to associate threat and issue contentiousness: the less an individual feels threatened personally by the actions of those with whom they disagree, the less they feel that the issue of hydraulic fracturing is contentious, and vice versa. The same pattern holds for threat against the United States; an individual’s threat tolerance is directly linked to their perception of issue contentiousness. The omega-squared score for personal threat is 0.081, indicating that perception of personal threat accounts for 8.1% of the variance in perceptions of issue contentiousness, while perception of threat towards the United States, although statistically significant, accounts for a much lower 1.8%.
Respondents were also asked to describe the collegiality of their relationships with people they disagree with on the issue of hydraulic fracturing. Responses options were again on a Likert-type four-point scale from “not collegial at all” to “completely collegial.” Average collegiality of relationships with those with whom a respondent disagrees was 2.23, around “somewhat collegial.” Considering these are relationships among policy actors who are on opposite sides of an issue, a low mean is a valid result. The association between relationships with those with whom you disagree and contentiousness is slightly convoluted but it is statistically significant. Respondents who take the extremes of collegiality, “not at all collegial” or “extremely collegial,” are associated strongly with, respectively, “more contentious” and “less contentious” perceptions on the issue of hydraulic fracturing. This is an intuitive result that if a
46


policy actor feels that their relationships are extremely collegial, even among those with whom they disagree, then it seems logical that the same policy actor would view the issue of hydraulic fracturing as less contentious. The patterns among the moderate responses on relationships do not display an obvious trend. The collegiality of relationships among policy actors who disagree accounts for nearly 4% of the variance in perception of issue contentiousness.
In regard to issue priority, there are strong and statistically significant patterns associated with perceptions of issue contentiousness, but in ways that do not conform to patterns observed in previous respondent cognitive characteristics. Respondents were asked: “How much of a priority is it for you professionally or personally to deal with political and policy issues related to oil and gas development that uses hydraulic fracturing?” Five response options ranged from “not a priority” to “the highest priority.” The mean was 3.69, above “a moderate priority” and close to “a high priority.” The sample for this survey was relatively highly-educated and more professionally than personally experienced on the issue of hydraulic fracturing, so this question may be picking up on that description. Nonetheless, there is a strong association that if an individual feels the policy issue of hydraulic fracturing is a high priority, they are also much more likely to perceive the issue as contentious, and the reverse holds for respondents who feel that the issue is of low priority.
Finally, respondents were asked to consider a measure of a policy outcome: protection of the environment and public health. Specifically, respondents were asked to gauge if protection of the environment and public health had become worse, stayed the same, or become better in the past two years. We might expect that those who believe that environmental and public health outcomes had deteriorated in the past two years would also perceive the issue to be more contentious. However, that expectation is not confirmed (see Figure 4.3). Although the
47


association between environment and public health outcomes and issue contentiousness is strongly statistically significant, discernable patterns are not readily evident.
Environment and Public Health
30
25 20 15 10 5 0
Less Contentious More Contentious
â–  Much Worse â–  Worse â–  About the Same â–  Better Much Better
Figure 3: Environmental and Public Health Outcomes by Contentiousness
If an individual perceives outcomes to be much worse, and to some extent, even much better, then they are also more likely to perceive the issue as less contentious. It may be that policy actors who perceive deteriorating or improving conditions also feel that an issue may be less contentious as conditions shift from status quo and change occurs, for better or worse. The average for this question was -0.05 (on a scale from -2 for “much worse” to 2 for “much better”); the majority of respondents feel that environment and public health outcomes are about the same now as they were two years ago.
Conclusion
The research presented here reviewed an in-depth case study of the perceptions and priorities of individual policy actors involved in oil and gas development that uses hydraulic fracturing. The policy conflict surrounding oil and gas development that uses hydraulic
48


fracturing is relatively contentious and there are several individual traits and cognitive characteristics that are associated with this position. Policy actors feel that outcomes, in regard to environment and public health, have slightly worsened over the past two years. The issue is a priority for these policy actors and they associate hydraulic fracturing with perceived threats, particularly personal threat perceptions. In terms of general traits, there are not statistically significant differences in education, stance, political position, and experience - with some caveats - that appear to drive an individual’s perception of issue contentiousness. Policy actors appear to be entrenched in their positions and threatened by the actions of those with whom they disagree.
For sustainable environmental policy, the implications of these patterns in cognitive characteristics are sizeable. Policy actor perceptions matter for making effective and sustainable policy that governs the extraction of oil and gas using hydraulic fracturing, while simultaneously protecting land and resources. Basic characteristics such as education, stance, or political ideology do not appear to be influential in determining issue contentiousness, but rather, the results of this study indicate that - in general - individual cognitive characteristics are more influential. Those who are threatened by the views of those with whom they disagree, have less collegial relationships with adversaries, and feel the issue is a high priority are also likely to be associated with a perception that hydraulic fracturing is a contentious issue. For policymakers and interested citizens, it appears to be important to moderate the levels of threat surrounding an issue and to encourage collegial relationships among opposing groups, in order to develop a policy area that is less contentious and consequently, more amenable to long-lasting and equitable solutions.
49


There are limitations to this study. As discussed earlier, any sampling procedure that attempts to identify a population that is not known apriori will encounter problems of sampling bias. We attempted to correct for that limitation by vetting the sampling procedure and final sample frame with issue experts, who felt that our population adequately represented the spectrum of individuals involved. A related limitation is that of selection bias. Respondents who chose to answer the survey may over-represent extreme positions as they might be more motivated to share their opinions owing to their strongly-held convictions on the issue. The rather moderate means on many questions would indicate that this is not an issue, but the risk remains and should be considered in drawing conclusions about a broader population of policy actors. Finally, any survey faces issues of survey instrumentality. The design and execution of a survey has been shown many times to be influential in responses. While we endeavor to present an unbiased survey instrument and to avoid all indications of preference, one respondent still felt that the survey was “misdirected” by taking up issues of hydraulic fracturing specifically when it should have been focused on shale oil more broadly. There is merit to this argument, as the policy issue of extraction of unconventional oil and gas has likely moved beyond just the technique of hydraulic fracturing, but the importance of studying this high-conflict policy issue remains high.
The potential impacts of hydraulic fracturing on land, water, air, and other natural resources are significant. Existing policy regulating hydraulic fracturing is fractured at best and lacks coherent best practices. Unlike some other environmental policies, the development of oil and gas extraction is an issue area that garners attention nationwide, from high-level lawmakers to everyday citizens. Those who live within proximity of industry operations inordinately bear the majority of the risks and harms associated with hydraulic fracturing, but the benefits are
50


diffuse and enjoyed by the entire population of the US, which presents significant policymaking challenges. The results of this study illuminate how individual traits and cognitive characteristics shape high-conflict policy issues. The data and analysis gathered here offer a rare insight into the personal belief structures of policy actors and how those belief structures shape issue perceptions. Theoretically, this study advances efforts to explain beliefs and preferences of policy actors involved in high-conflict policy issues. Practically, this study provides insight into how policy actors involved in high-conflict policy issues may differentiate themselves on perceptions of the level of contentiousness of a policy issue. Environmental policy is wrought with high-conflict issues, and it is imperative that we continue to explore how these issues take shape.
51


CHAPTER III
A CONFLICT WILD: EXPLAINING ACTIVITIES OF CONFLICT EXPANSION AND
CONTAINMENT
Understanding how people choose to participate in policy making processes is crucial to our understanding of policy outcomes, particularly in contentious policy debates. One of the primary political strategies in contentious policy debates is to expand or contain conflict (Schattschneider, 1960). In high-conflict policy debates, the strategy to expand conflict is important because more attention can be brought to the issue and more individuals can become involved, both of which can overcome the friction found in any political system. At the same time, for those wanting to defend the status quo or avoid the possible ramifications from the conflict, the strategy is conflict containment (Baumgartner, 1989; Pralle, 2006; Schattschneider, 1960).
Despite the ubiquity of these two strategies, within the policy process literature, relatively little is known connecting conflict expansion or containment with different forms of participation - also known as political behavior. Past research has emphasized two factors that can shape political behavior: sector (Brady et al. 1995; Carter 2016; Irvin and Stansbury 2004; Kam 2012) and individual-level beliefs (Mutz 2002, 2009; Olofsson et al. 2018; Weible and Heikkila 2017). This research considers how both sectoral affiliation and individual beliefs influence participation decisions regarding engaging in different activities to expand or contain the scope of conflict by asking: What drives conflict expansion or containment in contentious policy debates?
To understand how individuals make participation decisions in contentious policy settings, this research draws theoretical inspiration from policy process theories, specifically the
52


Advocacy Coalition Framework (ACF). As a theoretical framework, the ACF is appropriate for this study owing to its focus on the individual as the foundation of policymaking. According to the ACF, individuals form coalitions with the intent to impact the policy process, albeit in different ways (Sabatier & Jenkins-Smith, 1993). Individuals make intentional choices about how to influence policy over extended periods of time. Although the ACF recognizes the role of context as a modifier of political behavior (Weible & Ingold, 2018), empirical work most often focuses on the role of individual beliefs and disregards the influence of organizational-level context. Working under the ACF and drawing on the theoretical logic as found in Schattschneider and others, this research argues that individuals make strategic choices about their activities within the policy process to either expand or contain conflict. The study explores the different ways in which individuals choose to participate in a contentious policy debate, the use of hydraulic fracturing as an extractive technique for unconventional oil and natural gas.
The setting of this theory-testing case study (George & Bennett, 2005) is the use of hydraulic fracturing in Colorado. Hydraulic fracturing is a technique used to extract natural gas and oil from unconventional sources. The technique has grown markedly in its usage in the past two decades, which has been the source of contentious debate in the United States and worldwide (Evensen et al. 2017; Heikkila et al. 2014; Olofsson et al. 2018). The data for the study is taken from surveys collected in Colorado in 2015 and 2017. Individuals identified as active participants in the policy process were asked how they participate in the policy process and how effective those activities were in achieving their goals related to oil and gas development that uses hydraulic fracturing.
Although the survey sample was directed towards individuals ostensibly already involved in the policy process, or at least professionally attached to the issue, there were wide disparities
53


in the extent of participation among these individuals and their perceptions of the effectiveness of those activities. Factor analysis revealed distinct patterns of participation related to conflict expansion and containment, and ordinary least-squares regression was then used to uncover drivers associated with these patterns of participation. Results indicate that there is marked variation among patterns of participation, and that differences in participation patterns can be explained in part by differences in sectoral affiliation and individual beliefs.
Theoretical Background
Citizen participation is a vital part of a functioning democracy. In an open, transparent democracy, participation is crucial to claims of representativeness and accountability, particularly in a pluralistic political system. Much of the literature on participation considers political participation, such as voting for representation in elections. The focus of this research, however, is on what happens between elections via participation in public policymaking. Public policy, as defined by Dye, encompasses both government action and inaction - what governments choose to do as well as not to do (2002). Within public policy, participation can be influential in problem definition and solution development (Sobeck, 2003). Participation can provide input for decision makers; it can bring about learning and new information - perhaps even policy change - and is fundamental to democratic norms (Nohrstedt 2013; O’Faircheallaigh 2010).
Bishop and Davis undertook a thorough review of the extant literature surrounding participation in the policymaking process and concluded that the typologies and frameworks disjointed and in need of empirical testing (2002). They argued that theorizing about participation is “laden by idealist notions of democracy” and would benefit from clarity gained
54


through empirical research (2002: 14). Ten years prior in some of the first empirical studies on participation in the policy process, Roberts and King used grounded theory to explore how individuals active in the policy process might decide to participate, and if so, how (1991). We cannot understand the impact of participation without an awareness of how and why individuals participate in the policy process, yet there is a lack of solid empirical evidence that explains what influences participation differentiation among stakeholders. As Cairney notes, understanding the psychology of policymaking is critical as “to know why people make decisions, we need to know how they think before they act” (2016: 24-25). Knowledge of political participation in policymaking is inhibited by the under development of, one, the conceptualization of political participation and, two, theory explaining its variation. This paper contributes both.
Conceptualizing Political Participation: Expanding & Containing Conflict
Assuming that direct democracy or involvement in all government decisions is not possible for all citizens, Schattschneider argued that participation in policymaking is shaped by existing power structures (1960). Policy change is often the result of altering the scope of conflict; therefore, it is strategically important to expand and contract the scope of conflict within policymaking (Pralle, 2006). Schattschneider theorized that individuals, who are otherwise locked out of the policy process or lack enough power to change it, expand the scope of conflict using a variety of strategies to attract sympathizers and allies (1960). For example, in an attempt to “win”, these individuals might strategically act to broaden the discourse in the news media, recruit individuals new to the issue to support their side, and/or shift the issue to new decisionmaking venues.
55


Many of the frameworks and theories of the policy process pull theoretical insights from Schattschneider to address conflict expansion and containment (see Baumgartner & Jones 1993; Kingdon, 1984; Olson, 1965). The logic of Schattschneider is applied widely, but without adequate or encompassing conceptualization. The common focus has been on venue shifting as the mechanism with relatively less work on seeking new allies (see work by Baumgartner et al. 2009; Baumgartner & Jones 1993; Jones & Baumgartner 2012, for example). There is limited work in the ACF regarding conflict expansion or containment. Empirical work drawing upon participation in ACF research underemphasizes political engagement in that research most often explores participation as a determinant in the formation and maintenance of advocacy coalitions (Ingold and Varone 2012; Nohrstedt 2011; Weible et al. 2011) or as an explanation for policy change (c.f. Albright 2011). This research focuses on neither; rather, it focuses on the intermediary step of political engagement.
Pralle undertook one of the most comprehensive studies to date on participation via conflict expansion and containment in environmental policy settings, but focused solely on advocacy groups (2006). She recognized how the dueling roles of conflict expansion and containment shape the policy debate and outcomes. Her work revealed that the activities of individuals involved in policy advocacy are not random; they are the result of strategic and intentional choices made to influence policy. This paper adopts the basic conceptualization offered by Pralle and expands upon it by drawing inspiration from Goertz and Diehl (Goertz 2006; Goertz and Diehl 1993). In a seminal work on the constructs of enduring rivalries, Goertz and Diehl noted that there is a continuum of constructs, not absolute polarity (1993). Conflict expansion is similar in that we should expect shades of gradation when developing constructs of conflict expansion and containment via activity choice. We should see patterns of activity choice
56


among policy actors associated with goals of conflict expansion, or conversely, conflict containment.
There has been some empirical work that can generate limited expectations regarding the types of activities that are related to conflict containment or expansion. Past empirical work on conflict containment has demonstrated that important activities are related to countering arguments you disagree with and collaborating with the opposition to limit the scope of the conflict (Pralle, 2006). By dampening the conversation through either countering opponents or working with opponents surrounding a policy debate, the dominant coalition can maintain the status quo and contain the debate. Similarly, strategic brokering has been shown to be effective in containing conflict because it keeps the discussion within a single venue, thus benefitting the dominant, pro-status quo coalition (March & Olsen, 1984). Government agencies can also serve as a player in the policy processes, especially in implementing (hence supporting existing policies) or providing an opportunity for different political voices to be heard (e.g., through established administrative practices in rulemaking). With the government as a player itself in policymaking (Denhardt & Denhardt, 2000), sharing information and opinions with the government may also be important in maintaining the current direction of policy and, hence, containing the scope of the conflict. By commanding the conversation within the government, the dominant coalition can defend their jurisdiction (Hoberg & Phillips, 2011)
Mobilization has long been considered the primary activity for conflict expansion (Downs, 1957; Schattschneider, 1960). Schattschneider observed that increasing the number of allies in a coordinated effort can be significantly influential in expanding the scope of conflict (1960). Pralle established that “an important strategy for expanding participation in a conflict involves creating political alliances” (2006: 24). Baumgartner and Jones also found empirical
57


support for the tactic of coordinating allies with the intent to expand the conflict (1993). The media is often seen as a tool of conflict expansion, by changing the prevailing view of the issue (Baumgartner et al. 2009; Olofsson et al. 2017; Shanahan et al. 2011). Political participation constructed via activities individuals use to engage in the policy process is thus a continuum from containment-focused efforts, such as brokering or countering arguments, to efforts directed at expanding the scope of conflict, like mobilization or communicating with the media. Individuals may partake in some or all of these activities, at differing levels of effectiveness, in order to advance their goal of conflict expansion or containment.
Explaining Political Participation: Sector and Individual Influences
The field of public policy still has remnants of its behavioralist foundations by emphasizing the individual as the driver of political phenomenon. In doing so, however, the field overlooks the influence of organizational-level factors that enable and constrain political behavior. This paper incorporates both by identifying hypotheses that emphasize organizational-level factors (i.e., sector) as well as individual-level factors. Past research, albeit disjointed and in need of refinement, has shown that there are a multitude of ways in which individuals can choose to participate in either conflict expansion or containment, and those decisions are driven in part by sectoral affiliation and individual beliefs.
Sector
Under the banner of conflict expansion, Real-Dato undertook a detailed theoretical synergy of four major policy process frameworks regarding their respective interpretations of the influence of context as they relate to strategic participation in the policy process (2009). He asserted that context is a major determinant of participation decisions and called for empirical
58


applications exploring how context impacts stakeholder behavior within the policy process.
There are contextual nuances to participation that can inform decision making within the policy process. Irvin and Stansbury argue that context needs to be accounted for in order to design participatory practices that are appropriate for the community under consideration (2004). Context influences factors, both internal and external, that may impact opportunities and conditions for participation. Rydin and Pennington agree that participation in the policy process should be grounded by exploring the incentive structures of the involved individuals to better understand how and why those individuals choose to participate, and what the resulting outcomes might be (2011). Jones and McBeth claim that individuals make strategic choices about their activities in the policy process, with intent to influence the conflict (2010).
Individuals are motivated in different ways to participate in the policy process, and there is a lack of empirical research that identifies how those incentive structures might vary across contexts (Nohrstedt, 2013). Most research cites resources as the primary determinant in participation decisions (Hessels and Terjesen 2010; Pfeffer and Salancik 2003). Many scholars have called for participation studies to move beyond consideration of resources alone in determining participation decisions and to explore other ways in which participation decisions may be influenced (Brady et al., 1995; Irvin & Stansbury, 2004b).
Empirical research within the ACF literature recognizes context as a modifier of action, but typically at the level of political system (Fischer 2014; Leifeld 2013; Nohrstedt and Olofsson 2016; Sabatier 1998). In an effort to develop a framework of strategic political participation by private organizations, Hillman and Hitt discovered that “firm and institutional variables.. .affect the likelihood of making specific decisions” (1999: 825). Hillman and Hitt also found that participation was influenced differently by organizational-level characteristics than individual-
59


level characteristics. There has been some work at the organizational level (see Crow & Baysha, 2013; Weible & Sabatier, 2005), but there remains a lack of empirical studies. This research moves beyond political context to explore how organizational context as sectoral affiliation is related to participation decisions.
The limited work that has been done connecting activity choices in a contentious policy debate to conflict expansion has only shown that generally, individuals will undertake strategic action in order to meet their goals of conflict expansion or containment (Baumgartner and Jones 1993; Kingdon 1984; Schattschneider 1960). Individuals participate in the policy process according to whether they find themselves on the winning or losing side of the debate. In the state of Colorado, unconventional oil and gas extraction using the technique of hydraulic fracturing is currently occurring and has been for many years. The industry is comprised solely of companies in the private sector, who could be considered the winning side of the debate as their goal - extract resources using hydraulic fracturing - is currently allowed. Since extraction is happening, the industry and its partners dominate the question of whether hydraulic fracturing should be allowed. What remains now are largely questions related to how the extractive activity should be regulated. State and local governments provide the regulatory support for the activity to occur (Baumgartner & Jones, 1993). Following past empirical work on conflict expansion, it stands to reason, therefore, that private and public sector context will be associated with the dominant coalition and consequently, conflict containment.
Hypothesis la: Private and public sector will be associated with conflict containment.
There are some limited efforts to reverse Colorado’s status quo of allowing hydraulic fracturing to occur, but most effort on this side of the debate is focused on regulatory questions
60


surrounding how the process should be regulated. This side of the conflict could be considered the losing side, comprised almost exclusively of organizations and individuals in the nonprofit sector in Colorado. In an effort to expand the scope of the conflict, as past work on conflict expansion has shown to be the aim of the losing side of a policy debate (Baumgartner & Jones, 1993), individuals in the nonprofit sector, therefore, should be associated with patterns of conflict expansion.
Hypothesis lb: Nonprofit sector will be associated with conflict expansion.
Individual Beliefs
The ACF highlights the importance of beliefs as a uniting mechanism in the formation of coalitions (Fischer 2014; Sabatier and Weible 2007). Specifically, an individual’s beliefs regarding a policy issue will motivate their behavior to seek out other individuals who share the same beliefs and desire to participate in the policy process. The framework views the policy process as the interaction of competing advocacy coalitions, within which individuals share beliefs (Sabatier & Jenkins-Smith, 1993).
Extensive ACF empirical work has considered the role of beliefs within the policy process, as this assumption is fundamental to the framework (Fischer 2014; Ingold and Varone 2012; Weible 2005). Within the ACF literature, it is generally accepted that deep core beliefs are fairly resistant to change; there may be more flexibility - and hence opportunity for learning or policy change - within policy core beliefs and most certainly in secondary beliefs (Weible & Sabatier, 2006). It has been argued that the framework needs more empirical work on understanding how beliefs are differentiated, to what end that purpose serves, and the implications of beliefs on individual activity (Nohrstedt & Olofsson, 2016a). If beliefs and their
61


translation into policy are fundamental to coalition activity, it stands to reason that there should be a connection between beliefs and participation decisions. Foundational work from March and Olsen recognized early that “action depends on the decision maker’s values” (1984: 737). There is a multitude of conflicting evidence on the role of individual beliefs in influencing participation decisions. This leads to conflicting expectations. One of the goals of this research is to bring clarity to the empirical knowledge regarding the influence of individual beliefs on participation decisions.
In prevailing theory on conflict expansion, it has been demonstrated repeatedly that the dominant coalition in a policymaking process will prefer to contain the conflict and the nondominant coalition will endeavor to expand the scope of the conflict. In Colorado, where hydraulic fracturing is currently occurring, the dominant coalition should be comprised of individuals who hold a pro-stance on hydraulic fracturing and will be associated with conflict containment. This is hypothesis 2a. Relatedly, individuals who hold an anti-stance on hydraulic fracturing will be associated with conflict expansion. This is hypothesis 2b.
Hypothesis 2a: Pro-stance will be associated with conflict containment.
Hypothesis 2b: Anti-stance will be associated with conflict expansion.
Past research has shown that it is imperative that policy actors feel their actions may have a viable chance of influencing the process (Gray, 1989; Lubell, 2004; Sowa, 2009). A lack of viable options might lead to a lack of activity. Classic policy theory regarding conflict expansion focuses on the role of viable venues. Individuals who are on the winning side of a policy conflict are more likely to perceive readily available and receptive venues that are making decisions to maintain the status quo and contain the conflict. Those on the losing side may conversely feel a
62


lack of viable options and may pursue activities to expand the scope of conflict through alternative, currently non-viable venues instead (Holyoke et al. 2012; Pralle 2003). This leads to the third hypothesis, which argues that perceptions of viable venues will be associated with conflict containment.
Hypothesis 3: Positive perception of viable venues will be associated with conflict containment.
Finally, perceptions of threat can be influential in participation decisions (Cobb & Elder, 1972; Heikkila & Weible, 2017). Threat is often assumed to be an undercurrent of motivation for all policy actors in a given policy process. Schattschneider and Downs both recognized that threat is particularly useful for the losing side as it attempts to expand the scope of a conflict in the hunt for policy change (Downs, 1957; Schattschneider, 1960) but other work, particularly in the context of high-conflict policy debates, argues that both sides use threat (Jaspal & Nerlich, 2014). In the context of hydraulic fracturing, Jaspal and Nerlich demonstrated that threat is “deployed as a mobilizing factor on both sides of the debate - by pro-fracking and anti-fracking” (ibid.: 359). The final hypothesis predicts that increasing perceptions of threat will be motivating for both conflict expansion and conflict containment.
Hypothesis 4: Perceptions of threat will be associated with both conflict expansion and conflict containment.
Figure 1 illustrates the logic of the model of participation choice: sectoral affiliation and individual beliefs drive activity choice, which falls into patterns described by either conflict expansion or containment. This forms the dependent variable of the model: patterns of participation. There are several ways in which an individual can participate in the policy process,
63


and this research argues that 1) there are patterns to those activity choices and 2) the patterns conform to either conflict expansion or containment. Sector and individual beliefs are the independent variables of interest.
Figure 1: Determinants of Participation in the Policy Process Model Specification
The analysis draws upon a set of surveys first administered in Colorado in 2015 and repeated in 2017. The intent of the survey was to understand stakeholder perceptions of and priorities regarding the politics and regulations of hydraulic fracturing. Survey questions were inductively developed after in-depth interviews with key stakeholders and deductively driven by prior theory. The survey was reviewed in its entirety by an advisory board of influential and experienced individuals who are highly engaged in the topic of hydraulic fracturing. The sample
64


for the survey was purposive. Individuals directly or indirectly active in the policy debate were identified through internet research and snowball methods from interview responses and advisory board recommendations. These individuals are also known as “policy actors”. The survey sample is not intended to be generalizable nor a reflection of public opinion; it is intended to be a survey of the perceptions and attitudes of individuals who are already engaged in the policy debate to varying extents. The survey builds on previous survey work done in New York, Texas, and Colorado.
In total, 1004 people were administered the survey and 421 responded, for an overall response rate of 41.9%, which is acceptable for an online survey (Nulty, 2008). The data is a pooled cross-sectional set.7 In 2015, 453 people were administered the survey and 235 responded for a response rate of 51.9%. In 2017, 551 individuals were administered the survey and 186 responded for a response rate of 33.8%. The unit of analysis is the individual. Some predictors are measured at the organizational level, but these are considered to be influences upon the individual’s participation decisions and do not therefore necessitate a need for clustering.
Dependent Variable
The survey asked respondents to what extent they engage in and use effectively eight different activities to achieve their personal or profession goals related to oil and gas development that uses hydraulic fracturing: Brokering agreements between parties (Brokering), Countering arguments made by people you disagree with (Countering Arguments), Mobilizing the public (Mobilizing), Collaborating with people you disagree with (Collaborating),
7 The survey data is not a panel set. The sample was recreated each time to maximize the identification of individuals active in the issue of hydraulic fracturing in Colorado. A total of 64 individuals responded to both surveys.
65


Coordinating political activities with allies (Coordinating), Providing information to government officials {Government Information), Providing information to the news media {Media), Sharing your opinion with government officials {Sharing Opinion).8 These activities were vetted in interviews and with the advisory board to represent a comprehensive spectrum of participation. Responses were measured on a Likert-type scale from 0 (Not engaged) to 1 (Engaged, but not effectively) to 2 (Engaged and moderately effective) to 3 (Engaged and very effective).
In absolute terms, providing information to government officials was deemed to be the most effective activity for respondents to achieve their goals, followed by coordinating political activities with allies and brokering agreements between parties. Generally, if respondents were engaged in an activity, they reported moderate levels of effectiveness. Across responses and activities, the mean level of effectiveness was 1.45, between engaged but not effectively and engaged and moderately effective. Providing information to government officials had the highest average level of effectiveness (1.80), followed closely by sharing your opinion with government officials (1.72) and countering arguments made by people you disagree with (1.68). The average level of effectiveness for the remaining activities was much lower, with mobilizing the public reported to be the least average effective activity.
The effectiveness scale of the responses is key to conflict expansion and containment. As discussed in the previous section, developing constructs for conflict expansion and containment based on patterns of activity choice should be built upon a continuum, not a stark distinction (Goertz, 2006). For example, individuals hoping to expand the scope of conflict may participate in most if not all of these activities, but there is variation in the perceived levels of effectiveness.
8 See Appendix B for the full version of all relevant survey questions.
66


This is where factor analysis becomes most effective to discover latent constructs within the scaled responses.
Latent construct analysis through exploratory factor analysis (EFA) was used to determine if there are patterns among choice between activity types. Latent construct analysis aims to uncover underlying constructs among related items. It is often used in survey research when researchers want to summate or condense responses related to an overarching topic to reveal underlying patterns within responses (c.f. Finkel, 1985; Perry, 1996).9
Initial factor analysis revealed two primary factors.10 Orthogonal varimax rotation was used to distill the identified components. Best practices for break points in social sciences were followed (see Appendix C for Stata output of exploratory factor analysis).11 A scree plot of the eigenvalues of the factors against the mean was used to confirm the appropriate number of factors to keep was two (Cattell, 1966). The total variance explained between the two factors is 59.12%. Table 1 presents the factor loadings for each item.
Table 1: Rotated Factor Loadings by Item
ITEM ROTATED FACTOR LOADINGS
FI: “Containing” F2: “Expanding”
Brokering agreements between parties 0.7034
Countering arguments made by people you 0.6126
9 EFA can be used as an analytical tool in and of itself (Costello and Osborne 2005; Rummel 1970) or it can be used as an intermediary step in a larger analytical process, often to create independent or dependent variables that are more amenable to analysis (Santos, 1999). The latter approach is used in this research. Some researchers claim that using summated scales through latent construct analysis is more valid than using one response as an indicator of a broader concept (Rummel, 1970); however, many also note that use of latent construct analysis is a combination of science and art, simultaneously deductive and inductive.
10 Kaiser-Meyer-Olkin (KMO) is a measure of sampling adequacy, estimating the commonality among the items as appropriate for factor analysis. KMO measures for all items were above 0.70, which is acceptable for this research (Kaiser, 1974). The overall KMO score was 0.80, which is considered “meritorious” and more than sufficient for factor analysis.
11 Costello & Osborne note that due to the nature of social science data, traditional break points utilized in physical science research for retaining factors should be relaxed (2005). Therefore, two factors with eigenvalues above 1.0 were retained in this analysis and all components with rotated factor loadings above 0.60 were included (Kaiser, 1958).
67


disagree with
Collaborating with people you disagree with 0.6775
Providing information to government officials 0.7421
Sharing your opinion with government officials 0.7286
Mobilizing the public 0.8648
Coordinating political activities with allies 0.8648
Providing information to the news media 0.6973
The first factor - Containing - grouped Brokering agreements between parties,
Countering arguments made by people you disagree with, Collaborating with people you disagree with, Providing information to government officials, and Sharing your opinion with government officials. The eigenvalue for this factor was 3.50. This factor appears to capture policy actors working to contain or mitigate the scope of conflict by maintaining the status quo and conforms to past empirical work regarding patterns of activity choice related to conflict containment.
The second factor - Expanding - grouped Mobilizing the public, Coordinating political activities with allies, and Providing information to the news media. The eigenvalue for the second factor was 1.23. These policy actors appear to be more motivated towards expanding the scope of the conflict. These results conform to anecdotal observations of variation in type of participation and effectiveness among engaged policy actors and reveal that there is a distinct pattern to activity related to conflict containment as well.
Factor loading scores for each factor were then predicted for all observations. Factor scores represent a likelihood, from negative (less likely) to positive (more likely) of the respondent conforming to the response patterns of a given factor. The distribution centers around zero, with a standard deviation of 1. These scores comprise the dependent variable (continuous) for the OLS regression models.
68


Independent Variables
There are four independent variables: sector, stance, viability, and threat.12 The sectoral affiliation of each respondent was recorded as either public, private, or nonprofit. The sample includes 70 respondents in the nonprofit sector, 134 respondents in the private sector and 217 respondents in the public sector. Anti is a dichotomous variable that reports stance as the number of respondents who would prefer to completely stop or limit the use of hydraulic fracturing. About one-third of respondents reported an anti-stance on hydraulic fracturing. The reminder of the sample reported a pro-stance to continue at current levels or expand markedly the use of hydraulic fracturing.
Two independent variables capture the individual beliefs of respondents: Viability and Threat. Viability tests whether respondents feel that their actions to impact the policy process may have the potential in a receptive venue to actually achieve their goals related to hydraulic fracturing. For this dichotomous variable, respondents were asked whether there are any organizations or individuals who have the authority and trust to negotiate policy solutions to oil and gas issues in Colorado. 221 respondents, or nearly two-thirds of the sample, agreed that there are viable options to negotiate policy solutions.
Threat is measured in two ways. Respondents were asked to rank on a Likert-type, five-point scale the extent to which the views and actions of those with whom they disagree with on oil and gas development that uses hydraulic fracturing threaten them personally or professionally (.Personal Threat) or the state of Colorado {State Threat). The scale ranged from not at all (1) to a great deal (5). On average, feelings of personal threat were reported markedly lower at 2.95
12 See Appendix D for basic statistical description of the dataset.
69


compared with feelings of threat against the state of Colorado at 3.67. The distribution of responses for personal threat was much more evenly distributed than for threat against the state of Colorado, for which the responses were amassed at the far end of the scale.
Three control variables were used in the analysis. Education is a discrete nominal variable measuring the highest level of education attained. The sample is relatively highly educated; more than 70% of the sample reports at least a Master’s degree or higher. Respondents were asked how much of a Priority it is to deal professionally or personally with political and policy issues related to oil and gas development that uses hydraulic fracturing. The five-point, Likert-type scale ranged from not at all (1) to the highest priority (5). The mean was 3.77, indicating that this issue is a relatively high priority for most respondents. Finally, Year was included for year fixed effects to control for any within-year variation in the sample.13 First-order pairwise correlations did not reveal any concerns for collinearity among variables. The final model is:
Participation = Sector + Viability + Threat + Anti + Education + Priority + Year
Results
Table 2 presents descriptions for variables of interest by participation type: Containing versus Expanding. Means or modes on variables of interest were calculated to describe potential relationships among participation and sectoral affiliation and individual beliefs. Note that variables are measured on differing scales and are not continuous, therefore direct interpretation or comparison is not possible. However, comparisons among magnitudes are valid.
Table 2: Variable Description by Participation Type
13 There is no prior for year fixed effects to affect the DV.
70


Variable (scale) Factor 1: Containing Factor 2: Expanding
Sector (mode) Public Nonprofit
Stance (Pro/Anti) Pro Anti
Viability (0/1) Yes Yes
Personal Threat (1 to 5) 2.79 3.26
State Threat (1 to 5) 3.40 3.84
Education (Increasing) 5.38 4.91
Priority (1 to 5) 3.94 4.02
The mode is presented for sectoral affiliation. Among respondents who conform to the Containing type of participation, the most common sectoral affiliation is public and prevailing stance is pro. Among those respondents who can be characterized as Expanding participation type, the most common sectoral affiliation is nonprofit, but only marginally more so than private sector affiliation, and the prevailing stance is anti. Both types of participants believe that there are viable means for impacting the policy process. Threat, both personal and against the state of Colorado, is on average higher for Expanders but threat against the state of Colorado is higher than personal threat, regardless of participation type. Education is slightly lower among Expanders, and below the full sample mean as well. The issue appears to be a relatively similar priority for both types of participation. In summary, Expanders are primarily nonprofit sector actors who are slightly more threatened by the issue of hydraulic fracturing than their private and public sector counterparts, who are generally working to contain the scope of conflict.
71


OLS Regression Models
An OLS regression was run multiple times to determine if there are any significant relationships with sector and individual-level characteristics: first with factor loading scores for Factor 1: Containing as the dependent variable and again with factor loading scores for Factor 2: Expanding as the dependent variable, then separately with each individual activity component that feeds into the factors. Tables 3 and 4 present the results of the regressions with the factor followed by each of its components. 14
Table 3: Regression of Sector and Individual Characteristics on Factor 1: Containing and
Components14
Variables Factor 1: Containing Component: Brokering Component: Countering Arguments Component: Collaborating Component: Government Information Component: Sharing Opinion
SECTOR
Public 0.388*** (0.148) 0.045 (0.180) -0.234* (0.132) 0.150(0.151) 0.188 (0.137) 0.249* (0.131)
Private 0.207 (0.170) 0.042 (0.209) -0.015 (0.147) 0.187(0.164) 0.033 (0.158) 0.034 (0.155)
INDIVIDUAL
Viability 0.435*** 0.372*** 0.235** 0.335*** 0.391*** 0.375***
(0.112) (0.129) (0.094) (0.107) (0.109) (0.106)
Personal Threat -0.055 (0.047) -0.033 (0.055) -0.005 (0.043) -0.051 (0.045) -0.054 (0.046) -0.052 (0.043)
State Threat -0.014 (0.059) -0.001 (0.067) 0.019 (0.055) -0.071 (0.056) 0.071 (0.058) 0.085 (0.058)
Anti -0.366*** -0.328** -0.138(0.115) 0.046(0.119) -0.311*** -0.163 (0.119)
(0.122) (0.148) (0.114)
Education 0.106** (0.042) 0.081** (0.047) 0.086*** (0.033) 0.095** (0.042) 0.006 (0.043) 0.007 (0.040)
Priority 0.425*** 0.466*** 0.223*** 0.220*** 0.345*** 0.333***
(0.071) (0.077) (0.061) (0.066) (0.070) (0.072)
14 The component models were run as both an OLS and an ordered logit regression given the ordered nature of the dependent variable (from 0 = not engaged to 3 = engaged and very effective). Statistical significance was the same in both models. OLS coefficients are shown here to enhance interpretation.
72


Table 3 cont’d
2015 0.006 (0.102) 0.020 (0.119) -0.098 (0.088) -0.085 (0.103) 0.100 (0.099) 0.134(0.100)
Constant -2.335*** (0.439) -0.914** (0.443) 0.402 (0.393) 0.101 (0.407) 0.051 (0.425) -0.104 (0.409)
Observations 309 316 315 313 316 317
R-squared 0.240 0.171 0.161 0.111 0.154 0.130
Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.10
Table 4: Regression of Sector and Individual Characteristics on Factor 2: Expanding and
Components
Variables Factor 2: Expanding Component: Mobilizing Component: Coordinating Component: Media
SECTOR
Public -0.845*** -0.662*** -0.705*** -0.589***
(0.133) (0.163) (0.158) (0.143)
Private -0.478*** (0.149) -0.274 (0.181) -0.148 (0.173) -0.776*** (0.161)
INDIVIDUAL
Viability 0.137 (0.099) 0.277** (0.117) 0.233** (0.117) 0.189* (0.108)
Personal Threat 0.007 (0.044) 0.004 (0.048) -0.040 (0.055) 0.002 (0.049)
State Threat 0.144** (0.059) 0.132** (0.063) 0.245*** (0.070) 0.005 (0.063)
Anti 0.361*** (0.115) 0.489*** (0.137) 0.161 (0.131) 0.149 (0.130)
Education -0.062 (0.039) -0.127*** (0.043) -0.004 (0.048) 0.025 (0.041)
Priority 0.266*** (0.063) 0.235*** (0.082) 0.356*** (0.068) 0.361*** (0.069)
2015 -0.000 (0.096) -0.038 (0.109) 0.095 (0.113) -0.044 (0.106)
Constant -0.899** (0.368) 0.497 (0.434) -0.693 (0.443) 0.132 (0.429)
Observations 309 316 316 316
R-squared 0.336 0.263 0.292 0.205
Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.10
73


Sector appears to have a significant impact on participation decisions. For Containers, being in the public sector is positive and statistically significant, indicating that public sector respondents, all else equal, are more likely to participate as Containers than their counterparts in the private and nonprofit sectors. Further, although the variable Public is statistically significant for Expanders, the direction of the coefficient is negative, indicating that public-sector respondents are less likely to participate as Expanders, all else equal. The same holds for private-sector respondents; policy actors in the private sector are statistically significantly less likely to participate as Expanders, albeit at half the magnitude of effect than Public, compared to respondents in the nonprofit sector. In the context of hydraulic fracturing in Colorado, this is a confirmation of Hypothesis la - conflict containment is associated with the private and public sectors - and Hypothesis lb - conflict expansion is associated with the nonprofit sector. In addition, holding a pro-stance is statistically significant but negatively correlated with conflict expansion, and conversely an anti-stance is positive and statistically significantly correlated with conflict expansion. These results confirm hypotheses 2a and 2b.
In terms of individual characteristics, Viability was positive and statistically significantly related to Containing. Respondents who feel that there are viable options for impacting the policy process are increasingly likely to participate as Containers. The same did not find statistical significance for Expanding. This finding is intuitively valid. The activities that load on Containing - brokering agreements, countering arguments, collaborating across coalitions, informing government officials and sharing opinion with government officials - are activities that need viable venues or receptive decision makers in order to achieve success. The fact that viability is significant for Containers suggests that those mobilizing may feel locked out of the system and are invested in expanding the scope of the conflict to gain access to viable venues.
74


This is a confirmation of the third hypothesis.
Personal Threat was not statistically significant in either model. However, increasing perceptions of threat against the state of Colorado were positive and statistically significant, all else equal, for Expanders. Increasing perceptions of threat against the state of Colorado appear to be motivational in prompting activities that comprise Expanding: mobilizing the public, coordinating allies, and communicating with the media. This is a partial confirmation of the fourth hypothesis.
Component Analysis
Factor analysis is often criticized as blunt as a tool for measuring attitudes (Beavers et al., 2013). What do we pick up or loose when using factor analysis instead of using the components themselves? Best practices of factor analysis recommend comparing how correlations among factor scores compare to correlations among the components (Distefano, Zhu, & Mindrila, 2009). When considering each type of activity choice directly with sector and individual-level characteristics, some interesting insights appear.
OLS regression for Factor 1 indicated that being in the public sector was positive and statistically significant for individuals participating as Containers. However, component analysis revealed that public sector affiliation is only statistically significant for two of the five factors, and with a negative relationship for one of those factors (Countering Arguments). This indicates that policy actors who deem countering arguments to be a highly effective activity are not likely to be in the public sector. This runs counter to some prior work on conflict expansion in environmental policy settings (see Pralle, 2006).
Another departure from the results of the Factor 1: Containing is found within Anti
75


stance. Anti was found to be strongly significant but indirectly related to Containing, indicating that all else equal, individuals holding an anti-stance on hydraulic fracturing were less likely to participate as Containers and more likely to participate as Expanders. Again, though, only two of the five components display this same pattern: Brokering and Government Information. Anti was not statistically significantly associated with Countering Arguments, Collaborating, or Sharing Opinion. In the state of Colorado, it could be argued that those who hold a pro-stance on hydraulic fracturing are the current dominant group. Dominant groups generally prefer containment (Baumgartner and Jones 1993; Pralle 2009; Schattschneider 1960), and therefore the negative sign on the overall factor coefficient makes sense. However, there are some distinctions in terms of the components themselves.
The overall regression results from Factor 2: Expanding indicate that individuals in the nonprofit sector are most likely to participate as Expanders. As noted earlier, individuals in the private sector almost all hold a pro-stance on hydraulic fracturing and could be perceived as the dominant coalition in Colorado, where hydraulic fracturing is currently occurring. However, the strong indirect relationship between Private sector individuals and Expanding is driven entirely by the factor Media. Here the relationship is again negative, which makes intuitive sense that individuals in the private sector and on the dominant side are unlikely to seek communication with the media, and the same holds for individuals in the public sector.
The role of threat is an interesting finding across all components and factors. For those seeking to expand the scope of the conflict, State Threat is motivating to act, but in terms of individual factors, the relationship holds only to Mobilize or Coordinate. If participation in the policy process had instead been measured as one’s perception of effective communication with the media, which is common in numerous media studies surrounding hydraulic fracturing, State
76


Threat would not have appeared as a statistically significant relationship. This finding provides clarity to previous work on the role of threat in high-conflict policy debates. Much of that work asserts that powerful opposition by those with whom you disagree should be threatening and therefore motivation for action (Pralle, 2006). The results here demonstrate that threat may not be the motivator it is assumed to be, particularly when considering the goals of conflict expansion versus containment.
Conclusion
This research set out to understand variation in levels of participation in contentious policy debates by exploring the drivers of those patterns. There is a lack of research within the policy process literature that identifies how participation within the policy process may reflect the scope of conflict. The field lacks a solid conceptualization of conflict expansion as political engagement and relies too heavily on individual explanations, discounting organizational-level influences such as sector. This research fills that gap by first, conceptualizing conflict expansion via activity choice and second, explaining the drivers of activity choice as influenced by both organizational- and individual-level variables.
Findings indicate that there are specific activities related to either conflict expansion or containment, and individual choice among those activities is associated with sectoral affiliation and individual beliefs. This research is an important step in understanding which activities may be more effective or accessible to certain groups or policy goals. This is important in understanding structural barriers to participation and has implications for democratic participation and representativeness.
77


For organizational-level influences, the first hypothesis predicted that the private and public sectors would be associated with conflict containment and the nonprofit sector would be associated with conflict expansion. This was confirmed in the results. For individual-level influences, in Colorado, the dominant side currently holds a pro-stance on hydraulic fracturing and the opposition holds an anti-stance. The second hypothesis predicted that a pro-stance would be associated with conflict containment and an anti-stance would be associated with conflict expansion. This was also demonstrated in the results. The third hypothesis predicted that positive perceptions of viable venues through which to control the policy process would be associated with conflict containment, and this was confirmed. Finally, the fourth hypothesis was partially confirmed in that it expected to find beliefs of threat associated with both conflict expansion and containment, but findings revealed a statistically significant relationship between threat and conflict expansion only.
Too much of the public policy literature emphasizes individual factors over organizational. This paper shows both are important and shape the type of political activities people engage in (Weible & Carter, 2017). There are theoretical implications for the ACF in these findings. While much current work simply assumes that coalitions form and are maintained through shared beliefs, this research reveals that there is more going on. Individuals may be incentivized towards or constrained against participating, despite their beliefs, based on their context. This research corroborates much of what has long been suggested about the motivations for participation in the policy process: there is intendonality behind the activity choices that individuals make to influence policy, and this has implications for theory about coalitional politics in contentious policy debates.
In Colorado, the dominant side holds a pro-stance on hydraulic fracturing and is primarily
78


comprised of private and public sector individuals. The anti-stance side is currently largely comprised of individuals in the nonprofit sector. There is high collinearity in this particular study between stance and sector, and that is encouraging for the validity of the findings here. However, we can imagine circumstances in which the pro-stance would not be the dominant coalition, for example, settings with moratoriums against hydraulic fracturing like New York. In those places, the pro-stance coalition would work to expand the conflict and the results would be flipped: pro-stance should be associated with conflict expansion. We could also imagine a scenario in which there is not yet a dominant coalition - in new policy debates - in which case there might not be a significant relationship at all between stance and pattern of participation.
A reasonable question raised by this research is whether conflict expansion is possible in the public sector. The results here would indicate it is unlikely; however, remember that the survey question was not dichotomous - participate or not - but rather gauged the effectiveness of a given activity. Public sector policy actors may in fact engage in conflict expansion activities but feel that those activities are less effective in terms of conflict expansion. The institutional conditions of the public sector may prevent effective conflict expansion, and this remains a point for further research.
Relatedly, this research emphasizes the importance of developing a construct for understanding the relationship between activity choice and conflict expansion or containment. Relying on the past empirical work exploring specific activities in conflict expansion would suggest that measuring conflict expansion purely as “mobilization” would be a reasonable approach. For conflict containment, past findings would suggest that countering arguments made by persons with whom you disagree would be a sufficient measure of conflict containment. However, the results here demonstrate that this approach is not only insufficient, it actually
79


masks some of the more interesting nuances to individual-level determinants of participation.
There are limitations to this research that point to next steps in advancing research on conflict expansion and containment. First, using factor analysis to create the dependent variable dictates the nature and the distribution of the outcome measure. There are a limited number of applications that take this approach to first determine patterns within an outcome variable then use those patterns as a dependent variable (see Oser, Hooghe, & Marien, 2012). The non-ordered nature of nominal categories may call into question the choice of OLS regression as an estimation technique; ordering the types of activities by frequency, for example, might result in estimates that lend themselves to more intuitive interpretation. However, this approach masks the tendency of individuals to participate in regularized patterns and with differing values to different activities. An individual may only mobilize supporters once a year, registering as infrequent activity on many traditional measures of policy activity, but may feel that one effort is highly effective. The differentiation in activity weighting is captured through this research, which is a strength of the study.
Another limitation of this study is the selection bias of the sample, which constrains generalizability. While the study purposively sampled engaged stakeholders, it is surprising that there are many respondents who participate infrequently in the policy process. This could be an artefact of the survey design and the questions asked. Relatedly, another limitation of the study may be self-selection bias. There could be an overrepresentation of active participants who were more likely to answer the participation question, creating a systemic difference on the missing data observations. This requires further analysis to determine if the differences are indeed systematic and generating bias in the results or rather the result of the particular setting.
80


Finally, there may be barriers to participation in some activities. Some policy actors may perceive a lack of access in order to share their opinion or information with the government. Others may feel that they have no means of collaborating with those with whom they disagree or countering arguments made by those with whom they disagree. This, however, is one of the primary motivations for and contributions of this work: to uncover systemic inequalities in the system - perceived or actualized - and consider how those inequalities can be addressed.
The identification of distinct patterns of participation as conflict expanding or containing in nature is a solid contribution to the policy process literature. In addition, uncovering the influence of sectoral affiliation and individual beliefs on those patterns of participation is a first step towards better understanding how multiple contextual factors interact to influence participation decisions. More work remains to distill further influencers and continue to develop theories of participation in the policy process, so that we might better understand how contentious policy debates develop. The quality of political debate suffers when participation is marred by systemic obstacles. In the immortal words of Charlotte Bronte, “a conflict wild, and long and fierce the war will be” (1846).
81


CHAPTER IV
STRATEGIC MOTIVATIONS IN POLICY GOVERNANCE
Research in environmental governance has been generally characterized by collaboration studies. Most studies are done at the organizational level and focus on the structural characteristics of the network. Far fewer studies consider the individual network actor and how individual choices influence the resulting network. Most networks involve both choice and chance in formation. This research endeavors to explain some of the choice.
When making network choices, individuals face a selection problem. Owing to costs of networking and limited capacity for response from advocacy targets, individuals are forced to select partners strategically in a network space characterized by competition (Lee et al., 2012). There have been several studies of individual influence on strategic policy network choices (Fischer & Sciarini, 2013; Hadden & Jasny, 2017; Johari et al., 2006; Watts, 2001), but few studies consider how the context of policy actors or their individual perceptions influences their policy network choices. A theory of individual choice must account for two mechanisms: an external mechanism to account for the alternatives on offer and internal mechanism to account for choice between alternatives. Sectoral context as the external mechanism sets the menu of options available, and individual characteristics as the internal mechanism determine choice among those options. This research aims to uncover how individuals choose to build relationships by asking: How do sectoral context and individual perceptions influence the structure of policy networks? The analysis will model the incentives behind network choices and clarify network behavior by not just asking with whom an individual collaborates but also by assigning a value to that interaction. Exploring the determinants of how individual’s value relationships is a departure from many network studies and is a novel contribution.
82


This study focuses on policy actors, who are individuals identified as active in the policy issue and are distinct from the general public. The research assumes that policy actors make intentional choices regarding their network connections, and these choices are constrained by sectoral context and individual characteristics. The objective of this research is to identify variation in policy networks and how this variation can be explained by sectoral context and individual characteristics, through the lens of the Advocacy Coalition Framework (ACF). The ACF purports that policy issues are characterized by competing advocacy coalitions, which can be understood as networks of individuals (Matti & Sandstrom, 2013). This research will employ a relational approach (Ingold, 2011) to explore the policy networks of individuals involved in unconventional oil and gas development in Colorado. By adding an element of perception via importance, this research uncovers how individuals value relationships in networks. While the ACF recognizes the fundamental role of individual choice in network behavior, it lacks theory regarding how those network choices might be influenced by sectoral context and individual characteristics. The paper contributes to environmental governance and ACF literatures by interacting the sectoral context with individual characteristics to develop individual-level choice models explaining network preferences. The research will apply theories of organizational management, behavioral economics, and network science to assess the self-interested and benefit-maximizing choices of individuals on network structure.
Literature Review
Individual political behavior is constrained by time, financial resources, capacity, and most importantly, institutional rules and norms that vary by sector. Scharpf summarized the influence of institutional context on political behavior, noting that “institutions - a shorthand term for organizational capabilities and the rules governing their employment - will constrain,
83


but not completely determine, [political behavior]” (1989: 149). He goes on to distinguish between options that are “empirically infeasible” due to a lack of resources or skills and other options that will be “normatively prohibited” due to the institutional setting and its constituent rules. The institutional context - captured in this research as sectoral context - creates a setting within which some engagement options are impossible or at least severely disadvantaged. Sectors exhibit fundamentally different dynamics. The private sector can be characterized by economic opportunity, whereas the public sector may be characterized by “ constitutionally-sanctioned public choice” (Lynn, Heinrich, & Hill, 2001: 33). Governance arrangements are embedded in larger contexts and those contexts have implications for the execution of governance arrangements. For example, there are regulations and norms constraining the actions of nonprofit and public sector employees in regards to their efforts to impact the policy process. Recent work has also demonstrated that the influences on individual choices in relationship building are contingent upon context (Nohrstedt & Bodin, 2019).
Owing to a combination of sectoral context, individual cognitive limitations, and basic capacity constraints, policy actors are forced to make strategic decisions and target specific individuals or organizations to achieve their policy goals (Fischer & Sciarini, 2013). The policy networks that arise may be driven by these constrained decisions. Policy networks are the interactions of individuals and organizations with an aim to achieve personal or professional goals and can be viewed as the aggregation of many individual-level strategic choices (Gulati, 1999). Policy networks are constantly changing owing to contextual influences and policy actors’ network choices (Fischer & Sciarini, 2013). Policy actors may maximize their limited capacities by targeting those organizations and individuals whom they feel would have the highest potential impact on the policy process.
84


Several scholars recognize the importance of policy networks for power, legitimacy, resources, information, and advocacy (Agranoff & McGuire, 2001; Hill & Lynn, 2005; Lubell, Schneider, Scholz, & Mete, 2002; Milward & Provan, 2000; O’Toole & Meier, 2014; Provan & Milward, 2001). Much of the empirical policy network research focuses on collaborative networks under the guise of governance or collective action studies. Governance networks arise as a means of alternate service delivery (Ansell & Gash, 2007; Jing & Besharov, 2014; Provan & Milward, 2001; Wood, Donna & Gray, 1991), more inclusive government (Scott & Thomas,
2017), dealing with complexity (O’Toole, 1997; Wood, Donna & Gray, 1991), and generating positive outcomes (Provan & Kenis, 2008; Provan & Milward, 2001). These studies have focused on the structural characteristics of the networks and how networks are determinants of outcomes. Governance studies often recognize the role of constraints and competition within an advocacy space (see Lee, Lee, & Feiock, 2012), but fail to advance a full network theory of the policy process as it fails to explain the characteristics of the components of a network - the individual (Dowding, 1995).
There appears to be some level of strategic awareness among individuals and/or organizations of the importance of intentional decision making regarding networking building. Lee et al. explored the structural features of an interorganizational network to answer questions surrounding the emergence of policy networks, the forces that shape network formation, and how networks evolve over time (2012). The authors found that organizations seek to build stable and sustainable relationships, which require more than simple exchange transactions. Past work in policy networks, particularly in regards to policy change, have shown that network partners are at least somewhat aware of the advantages to be gained through strategic tie building (Chalmers, 2013; Granovetter, 1985). Leifeld and Schneider recognized the role of political opportunity
85


structures and transaction cost analysis in individual decision making regarding whom policy actors choose as contacts (Leifeld & Schneider, 2012). Using the case of information exchange, the authors determined that policy actors knowingly attempt to minimize transaction costs and maximize returns within their constrained capacity. This paper builds upon this finding by capturing the strategic choices of policy actors in network connections, as influenced by sectoral context and individual characteristics.
This research also moves a step beyond previous network studies by exploring not only with whom a policy actor chooses to connect but also the level of importance that policy actor assigns the interaction. By giving a measure of importance, the policy actors reveal how they value that relationship by reporting the utility of that interaction, which can yield deeper insight about strategic network choices. We learn more about relationship building in governance settings. With the individual as the unit of analysis, as opposed to the majority of studies done at the organizational level, this research can explore the institutional opportunities, constraints, and incentives in combination with individual characteristics that influence strategic network choices. Lee et al. remarked that specifically “in the study of public administration and policy, extant research on (interorganizational) networks has been focused on how networks work and how networks help individuals or organizations perform better based on macrolevel explanation.. .however, less attention has been directed to microlevel analysis of the emergence and structural features of interorganizational networks” (2012: 567-568). The authors note that additional empirical work could answer questions such as: “What are the underlying forces that shape the observed network configurations?” (ibid.). The field of network science has recognized that “instead of trying to find micro-processes that lead to certain network properties.. .one could follow a different analytical strategy and try to come up with analytical strategies that match with
86


actual behavior” (Snijders, Matzat, & Reips, 2012: 2). This current project endeavors to lend insight this empirical puzzle.
Conceptual Framework
A core assumption of the ACF is that individuals come together in coalitions to influence policy (Weible, Sabatier, & McQueen, 2009). The ACF argues that individuals hold beliefs regarding policy; those beliefs have been shown to be influential in motivating coalition development (see Henry, Ingold, Nohrstedt, & Weible, 2014a; Matti & Sandstrom, 2013; Weible, 2005). Coalitions are often identified based on “the assumption that common beliefs are reflected in relations among actors involved in policy processes” (Ingold, 2011: 435). It remains unclear, however, whether those beliefs alone are sufficient to bond advocacy coalitions (Henry et al., 2014b; Schlager, 1995), and significant empirical work has identified networks as a potential tool to better understand coalitions and coalition development (Fischer et al., 2012; Ingold, 2011; Lienert et al., 2013). This research does not dispute that shared beliefs can encourage the formation of policy coalitions as networks but rather aims to bring more theoretical clarity to that assumption in the ACF by testing whether tie formation might be motivated by context as well. For example, there is substantial empirical evidence that belief homophily is a driver of network formation (Jenkins-Smith et al., 2014; Metz, Leifeld, & Ingold,
2018); this research seeks to explore how other contextual variables, specifically sectoral affiliation, may also influence relationship building.
The goal of this research is to identify and explore political behavior as networks that reflect whom an individual policy actor targets to achieve his or her policy goals. These are related to but may also be distinct from policy coalitions. This study recognizes that coalitions exist as part of a larger policy network but is not explicitly concerned with coalition
87


identification. The policy networks in this study are not intended to represent or define coalitions15, but rather will be differentiated by sectoral context and individual characteristics.16 Network science has shown that there is much to be gained theoretically by considering network structure, moving beyond position- or belief-driven coalition structure (Bodin, 2017; Jackson, 2005). In this study, policy actors holding opposing stance on a policy issue may find themselves in the same policy network when conditioned on institutional context such as sector. The ACF is a framework that enables multiple theoretical arguments (Jenkins-Smith et al., 2014), and while the goals of this research are somewhat unusual within the framework’s empirical work, it still conforms to ACF assumptions. There is overlap in this question compared to what is traditionally asked in ACF coalition studies. This research expands knowledge regarding network relationships, supplementing theoretical and empirical ACF findings with knowledge gained in behavioral economics, organizational management, and network science.
This study models the incentives to foster valuable relationships and brings clarity to network behavior. By focusing on the individual instead of the organization, the research reveals a different perspective on governance and how individuals value relationships in policy networks. Early empirical work within the ACF by Zafonte and Sabatier found that although beliefs were important in understanding coalitions, beliefs alone were not enough to explain coordination (Zafonte & Sabatier, 1998). Scholars have continued to build on this foundation recognizing that there are interdependencies between network structure and belief systems (see Henry, Lubell, & McCoy, 2011; Schneider, Scholz, Lubell, Mindruta, & Edwardsen, 2003;
15 There are several reasons why this particular network analysis does not intend to define nor represent coalitions. First, the network is unidirectional and not symmetrical, in that the ties are not necessarily reciprocal. Second, the network is also bipartite, or two-mode, composed of two different types of vertices, in this case individual-organization. Finally and relatedly, the network is incomplete, as is common in social network analysis.
16 It should be acknowledged that some network studies have defined and identified coalitions using the type of network described proposed here. There is a sizeable literature within the ACF that debates how coalitions should be defined: by beliefs, or by network connections, or by shared venues, among others.
88


Weible, 2005). While there has been significant work exploring the connection between belief systems and coalition structure, there remains a gap in ACF literature regarding network valuation, specifically tie valuation. First, the focus has been on structural features of the network, such as transitivity or reciprocity (cf. Scott & Thomas, 2017), and second, extant literature has failed to take into account how context might interact with belief systems to influence network structure. Leifeld and Schneider found that the context of the target organization can impact the network choices of policy actors (2012). In 2011, Ingold built upon previous studies recognizing the role of individual beliefs in networks by studying if cooperative beliefs result in collaboration (Ingold, 2011). The author argued that although beliefs were shown to be instrumental in predicting network cohesion, the settings may constrain or enable network ties as well.
Homophily has been found to be a strong determinant of ties (Calanni et al., 2015; Lee et al., 2012; Leifeld & Schneider, 2012; Mcpherson, Smith-Lovin, & Cook, 2001). However, this stream of research focuses on beliefs as the driving force of homophily and often misattributes the social mechanisms underlying network structure to social influence when strategic partner selection may be more impactful (Lewis, Gonzalez, & Kaufman, 2011; Steglich, Snijders, & Pearson, 2010). The approach in this study recognizing the strategic tie behavior of policy actors extends the sphere of homophily to the sectoral context, which lends insight into how opposing actors may engage in similar tie-building strategies.
When tie behavior on networks is intentional, as it arguably is in policy advocacy networks, then we can turn to organizational management and network science for empirical direction on expectations, complemented by behavioral economics. Theories of behavioral economics adopt tools of economic analysis to explore, for example, political issues. Behavioral
89


economics assumes that individuals act strategically and in their self-interest to maximize their utility (Downs, 1957; B. D. Jones, 2001; Shepsle, 2008). These assumptions are not limited to economic choices; this applies political behavior in choice of network targets as well (Currarini, Jackson, Pin, & Papadimitriou, 2010). Revealed preference theory in economics suggests that when a consumer chooses one product over another, we learn something about that individual’s decision-making structures (Sen, 1971). Similarly in a network context, we learn something about an individual’s decision-making structures when he or she reveals their preferences through choice among target organizations in an advocacy setting. Theoretical economic models of individual strategic choice in network structure have been developed (see Bloch & Jackson, 2006; Galeotti, Goyal, & Kamphorst, 2006; Goyal & Vega-Redondo, 2005; Jackson & Watts, 2002; Jackson & Wolinsky, 1996) but empirical applications are lacking.
Individuals have limited relationship budgets, as Burt argues in his network theory of social capital (2000). Forming and maintaining network ties carries considerable uncertainty and transaction costs (Lee et al., 2012) and this uncertainty is linked to context (Nohrstedt & Bodin,
2019). In empirical applications, the cost and associated risk of forming links have been shown to be a major determinant in strategic network formation (Berardo, 2014; Goyal & Vega-Redondo, 2005). As the cost of forming links rises, fewer links will be formed. In addition, under conditions of constraint, individuals who perceive a policy issue to be contentious appear to be motivated form what they hope will be high-quality links (Ansell & Gash, 2007; Gray, 1989). Issue contentiousness seems to play a role in individual decision making regarding relationship building, but that role is not identical across sectors. If we assume that the cost of building relationships is relatively similar across sectors, then it stands to reason that issue contentiousness will play a larger role in relationship building in sectors with fewer resources,
90


such as the nonprofit or public sectors (Hessels & Terjesen, 2010; Miller-Millesen, 2003). Guo and Acar identified the conditions under which nonprofit organizations are more likely to collaborate and found that resources are a major determinant in network structure (2005). The authors emphasize, however, that resource dependency and transaction cost theories alone are not sufficient explanations for strategic network choice - context matters as well.
HI: Perceptions of issue contentiousness will be relatively more influential in relationship building among nonprofit policy actors than among public- and private-sector policy actors.
Past empirical work has shown that an individual’s perception of success owing their network efforts is influential in their determination of whether a relationship will be formed and maintained - subsequently referred to as viability (Ansell & Gash, 2007; Bradford, 1998; Geoghegan & Renard, 2002; Rogers et al., 1993; M. Schneider et al., 2003). Extensive work in network science has considered how to build a viable network and the traits that encourage network efficiency and stability, yet there remains a gap in knowledge at the individual level that explores how perceptions of viability influence an individual’s network choices, specifically under different sectoral contexts. Collaborative governance scholars recognize that inter-sectoral networks are crucial to the delivery of public services (Hill & Lynn, 2005; Milward & Provan, 2000; Provan & Milward, 2001). Foundational work in collaborative governance established that perceptions of viability are integral to public sector organizations embedded in governance arrangements (Provan & Milward, 2001). This research inherently recognizes that there are differences in sectoral affiliation among network participants but has not yet directly addressed how those sectoral differences may enable or constrain network choices. Past work in nonprofit research has shown that before even engaging in network activity, nonprofit organizations first consider potential for benefits (Nicholson-Crotty, 2009). Sowa demonstrated that perceptions of
91


Full Text

PAGE 1

DECONSTRUCTING POLITICAL BEHAVIOR: THE INFLUENCE OF CONTEXT AND INDIVIDUAL CHARACTERISTICS by KRISTIN L. OLOFSSON B.A., University of Colorado, 2001 M.Sc., Uppsala University, 2012 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 2019

PAGE 2

ii 2019 KRISTIN L. OLOFSSON ALL RIGHTS RESERVED

PAGE 3

iii This dissertation for the Doctor of Philosophy degree by Kristin L. Olofsson has been approved by Public Affairs Program by Christopher M. Weible, Chair Tanya Heikkila William Swann Daniel Nohrstedt Date: May 18 , 2019

PAGE 4

iv Olofsson, Kristin L. ( PhD, Public Affairs ) Deconstructing Political Behavior: The Influence of Context and Individual Characteristics Thesis directed by Professor Christopher M. Weible ABSTRACT People, policy, and power are inextricably linked. There is wi de variation in how individuals access and are able to influence policy, which is disheartening for the practice of democracy and fair and equal representation. This research studies the execution of power within policy by exploring how people participate in policymaking, as evidenced through their political behavior. Political behavior comes in many forms, such as writing a letter to the editor, protesting, or lobbying. Current research might be able to tell us who is likely to participate or even when th ey might participate but it does not explain how that individual made his or her participation decision and in what ways that decision was enabled or constrained. This research explores the determinants of political behavior decisions by focusing on two as pects: context and individual characteristics. Taking an institutional view of decision making, the research argues that external context, recognized here as sectoral affiliation, and internal mechanisms, captured through individual characteristics, intera ct in detectable and predictable ways to manifest in patterned political behavior. Results indicate three main themes. First, there is polarization in belief extremity related to perceptions of issue contentiousness. As an individual perceives an issue to be increasingly more contentious, they are also likely to associate heightened feelings of threat and a lack of viable solutions. Second, political behavior choices are not random; there is an element of intentionality or strategy behind the choices. Final ly, context appears to amplify some individual characteristics and dampen others. The research bolsters explanatory capacity of

PAGE 5

v theories of policymaking and contributes to practical knowledge about how we can better understand our setting in order to impro ve the quality of political discourse. The form and context of this abstract are approved. I recommend its publication. Approved: Christopher M. Weible

PAGE 6

vi DEDICATION This dissertation is dedicated to all the patience and hard work of my advisors , family, and friends. To Chris, Tanya, Will, Daniel, and many other mentors, your guidance throughout this process has been invaluable. I will be forever grateful for everything that you hav e done , and I hope to continue the tradition with my own students. To my parents Mike and Sylvia , you were my first and best teachers. You taught me everything important. To Koren and MJ , thank you for putting up with me and always being there, without judgement but usually with some sarcasm. I need someone to keep me in line. To Rolf and Ulla, your support and curiosity has been such an encouragement throughout this process. You are always there with a smile and a willingness to listen, for which I cannot thank you enough. To my amazing children Emmelina and Stellan , you are my source of inspiration. Having you has made every bit of this better. I love you more than you will ever know. And finally, to my amazing husband Robert , words cannot express how much you mean to me. been there through the good and bad, and never wavered. You are my best friend and my favorite person. You really deserve all the credit.

PAGE 7

vii TABLE OF CONTENTS I. INTRODUCTION ................................ ................................ ................................ ....................... 1 Background Literature on Political Behavior ................................ ................................ ............ 7 Conceptual Framework ................................ ................................ ................................ ............ 12 Objective 1: Role of Individual Characteristics ................................ ................................ ... 15 Objective 2: Activities ................................ ................................ ................................ .......... 15 Objective 3: Policy Networks ................................ ................................ ............................... 17 Research Design ................................ ................................ ................................ ....................... 22 Case Selection ................................ ................................ ................................ ...................... 23 Methods of Data Collection ................................ ................................ ................................ . 25 Sampling: Policy Actor Identification ................................ ................................ .............. 25 Key Variable Operationalization ................................ ................................ ...................... 28 Analysis ................................ ................................ ................................ ................................ .... 29 Limitations ................................ ................................ ................................ ............................... 31 Contributions ................................ ................................ ................................ ............................ 32 II. PERCEPTIONS OF CONTENTIOUSNESS: HOW INDIVIDUAL TRAITS SHAPE ENVIRONMENTAL POLICY CONFLICTS ................................ ................................ .............. 34 Case Setting: Hydraulic Fracturing ................................ ................................ .......................... 36 Data Collection ................................ ................................ ................................ ........................ 38 Results ................................ ................................ ................................ ................................ ...... 40 Respondent Belief Structures ................................ ................................ ............................... 41

PAGE 8

viii Conclusion ................................ ................................ ................................ ............................... 48 III. A CONFLICT WILD: EXPLAINING ACTIVITIES OF CONFLICT EXPANSION AND CONTAINMENT ................................ ................................ ................................ ......................... 52 Theoretical Background ................................ ................................ ................................ ........... 54 Conceptualizing Political Participation: Expanding & Containing Conflict ........................ 55 Explaining Political Participation: Sector and Individual Influences ................................ .. 58 Sector ................................ ................................ ................................ ................................ 58 Individual Beliefs ................................ ................................ ................................ .............. 61 Model Specification ................................ ................................ ................................ ................. 64 Dependent Variable ................................ ................................ ................................ .............. 65 Independent Variables ................................ ................................ ................................ .......... 69 Results ................................ ................................ ................................ ................................ ...... 70 OLS Regression Models ................................ ................................ ................................ ....... 72 Component Analysis ................................ ................................ ................................ ............ 75 Conclusion ................................ ................................ ................................ ............................... 77 IV. STRATEGIC MOTIVATIONS IN POLICY GOVERNANCE ................................ ............ 82 Literature Review ................................ ................................ ................................ ..................... 83 Conceptual Framework ................................ ................................ ................................ ............ 87 Methodology ................................ ................................ ................................ ............................ 93 Results ................................ ................................ ................................ ................................ ...... 96 Exponential Random Graph Model ................................ ................................ .................... 102

PAGE 9

ix Conclusion ................................ ................................ ................................ ............................. 106 V. CONCLUSION ................................ ................................ ................................ ...................... 111 What We Learned ................................ ................................ ................................ .................. 115 Limitations ................................ ................................ ................................ ............................. 117 Why It Matters ................................ ................................ ................................ ....................... 119 REFERENCES ................................ ................................ ................................ ........................... 120 APPENDIX A: Concepts and Operationalization ................................ ................................ ........................... 135 B: Full Survey ................................ ................................ ................................ ............................. 136 C: Factor Analysis (Principle Components with Orthogonal Varimax Rotation) ...................... 144 D: Basic Statistical Desc ription of Dataset ................................ ................................ ................. 145

PAGE 10

1 C HAPTER I I NTRODUCTION An essential part of the policy process is the maneuvering of political behavior to impact policy change and outcomes . Political behavior comes in many different forms , such as lobbying, writing letters to the editor, or sharing your opinion with government agencies . The development of a special task force in the state of Colorado to deal with policy i ssues related to oil and gas regulations illustrates many aspects of political behavior. While Colorado has been involved in oil and gas extraction for many years, the issue has historically been contentious, particularly in recent years with the increased use of hydraulic fracturing as an extractive technique. As this highly contentious issue has progressed, many individuals have been involved in a myriad of ways in attempting to influence the policy surrounding hydraulic fracturing. Some of the ways in wh ich individuals have attempted to influence policy are creating citizen groups, joining environmental groups, testifying at public hearings, advocating with influential decision makers, and sharing their opinion in the media, among several others. One exam ple of a particularly visible way of participating in the policymaking process was the creation and implementation of a task force by then Governor John Hickenlooper and former House Representative Jared Polis (now governor) to address a volatile situation that arose in 2014. Stemming from 2012 when the city of Longmont in Northern Colorado passed a ban on hydraulic fracturing, the task force was essentially a compromise between existing pro and anti coalitions to avoid a ballot issue and a contentious law suit involving the city of Longmont and the Colorado State regulatory authority, the Colorado Oil and Gas Conservation Commission (COGCC) (Jaffe, 2014) .

PAGE 11

2 The task force was comprised of an assortment of representatives from industry, local government , and environmental and other interests. It was one of the first collaborative, inter coalitional venues available in the highly conflictual policy issue of hydraulic fracturing in the state of Colorado. 1 The creation of an inter coal itional task force trig gered a movement towards a less adversarial strategy from both coalitions. In this highly visible venue, individuals worked together to respond to a shifting landscape of public opinion and demands for transparency and accountability from all sides. The actions of the individuals that brought about the creation of the task force varied, including but not limited to media campaigns, research production and dissemination, lobbying , and testifying at public hearings. There were numerous venues in which t hese actions occurred, and the networks of individuals involved were diverse , from federal to local level government officials to community nonprofits to the general public. The coalesce nce of activities and policy networks into political behavior prompted Governor Hickenlooper to respond, in what eventually had a sizeable impact on the November 2014 ballot issues in Colorado. Current theory in political behavior might be able to predict when an individual participate s and perhaps even why that individual chooses to participate, but existing theory does not tell us how an individual made that political behavior decision. We cannot understand the impact of political behavior within the policy process without understandin g how individuals make decisions. To better understand individual political behavior within the policy process, this dissertation focuses on two of its aspects: activities and policy networks . Activities are defined as the actions taken by policy actors to influence the policy process , such as the lobbying efforts of 1 gas issues slated for the November 2014 ballot were dropped. Two initiatives were from the anti drilling coalition and two were from the oil and gas industry and its allies.

PAGE 12

3 the individuals involved in the Colorado case . Policy n etworks represent the interactions of individuals and organizations with an aim to achieve personal or professional goals. This political behavior is influenced by two general factors: context and individual characteristics. Institutional context is comprised of the rules and norms of the broader setting in which an individual is operating, such as the public, private, or nonprofit sector. The institutional context varies the rules of the game in different settings. Individual characteristics are defined as personal belief systems or cognitive tendencies, such as the tendency to hold strong beliefs risk perceptions or determination of issue salience. Figure 1 displays the relationship between these concepts, ultimately producing political behavior. Figure 1: Model of Political Behavior This dissertation uses a multi method case approach to expla in how institutional context and individual characteristics interact to modif y political behavior in a high conflict policy issue . This research abandons when questions of political behavior and instead explores how and why institutional conditions and individual characteristics interact to const rain or enable that behavior by asking: How do institutional context and individual characteristics influence political

PAGE 13

4 behavior in contentious policy debates? The dissertation views political beha vior through the theoretical lens of the Advocacy Coalition Framework (ACF) and supplements from several theories such as institutionalism, policy networks , conflict expansion, and behavioral economics to develop t hree chapters that will explore different aspects of how context modifies political behavior. Chapter II: Role of Indivi dual Characteristics . Objective: To describe how individual characteristics shape environmental policy conflicts . Understanding the policy process requires understanding the individuals involved in environmental policy making. This objective will consider how the beliefs and policy preferences of individuals involved in the policies shape the extraction of unconventi onal oil and gas in the US. Data for this foundational chapter is drawn from a 2016 survey and associated interviews of individuals involved in hydraulic fracturing at the national level. Chapter III : Act i vities . Objective: To identify patterns in the typ es of activities used by policy actors in the policy process and how these patterns are influenced by institutional context and individual characteristics . According to the ACF, individuals participate in the policy process, via coalitions, with intent to influence the policy process (Sabatier & Jenkins Smith, 1993) . This objective explores how institutional context, such as sector, may interact with individual level characteristics, such as strength of beliefs, to encourage or thwart individual activity within the policy process. This chapter dives into the context of Colorado, using data f rom a set of surveys and associated interviews of individuals involved in hydraulic fracturing in the state. The survey was first done in 2015 and repeated in 2017. Chapter IV : Networks . Objective: To identify variation in policy networks and how this vari ation can be explained by differences in institutional context and individual characteristics .

PAGE 14

5 The ACF purports that policy issues are characterized by competing advocacy coalitions, which can be understood as networks of individuals (Matti & Sandström, 2013) . This objective employs a relational approach (Ingold, 2011) to explore the policy networks of individuals involved in unconventional oil and gas development in Colorado. While the ACF recognizes the fundamental role of individual choice in network behavior, it lacks theory regardi ng how those network choices might be influenced by institutional context and individual characteristics. This objective applies theories of behavioral economics and strategic network formation to assess the self interested and benefit maximizing choices o f individuals on network structure , using the same Colorado dataset as in the previous chapter . By focusing on the influences of political behavior decisions , t he dissertation generates k nowledge about the diversity of political behavior that might be used in different contexts and contributes to our understanding of the policy process and political outcomes. The theoretical definition of political behavior can be refined by exploring the use of activities and policy networks at an individual level. Exploring distinct institutional settings with varying options for activities and political networks can lend insight into why certain political behaviors emerge. Individuals filter institutiona l context through their individual characteristics , and that influences their political behavior. We can detect institutional settings and individual cognitive characteristics that shape political behavior. In addition, t his research has implications for u nderstanding contentious political exchanges. If we better understand the different toolbox that each person has to interact in the policy process, then we can tackle difficult issues such as disproportionate representation and unequal and inequitable outcomes. Democracy is built upon a fundamental tenet that each person if they so desire should have a voice in the process. The

PAGE 15

6 dissertation proposed here aim s to clarify how people choose to exercise that voice and what determines those individual decisions. The setting for the research is the exploration of shale oil and g as through hydraulic fracturing . Hydraulic fracturing is a technique used to extract na tural gas and oil from unconventional sources. To reach unconventional sources of oil and natural gas, vertical wells are first drilled and then a horizontal well is drilled to the source point. Hydraulic fracturing fluids are forced into the wells at high pressure in order to fracture the rock and prop open the fractures after breakage. The released resource can then flow through the horizontal and then the vertical wells for extraction. With advancements in horizontal drilling, t he technique has grown mar kedly in its usage in the past two decades, which has been the source of extensive debate in the United States and worldwide (Heikkila, Pierce, et al., 2014; Rahm, 2011) . Issues surrounding hydraulic fracturing include air and water quality, public nuisance, energy security, usage of public and private land, mineral rights, siting of wells, and impa cts on the economy. T he research first approaches political behavior within the policy process by exploring the general associations between individual characteristics and policy conflicts. This work is done at the national scale, to avoid possible issues of contextual variation. However, the primary interest of the dissertation is precisely contextual variation, which is addressed through the case study in Colorado. The state is a representative setting in which to explore political behavior within the policy process as the state has a long history with oil and gas development and is accustomed to issues associated with this particular industry, while opposition remains visible an d vocal with a diversity of actors in a diversity of institutional settings . The study population for this research is not the public but rather individuals who are already involved in the policy process. These individuals are known as policy actors. In th e earlier example , policy actors were

PAGE 16

7 those individuals who were involved in the debates that led to the creation of the task force. Drawing upon survey s and interviews in Colorado in 2015, and 2017, policy actors were asked how they participate in the pol icy p rocess and to what extent and to ward whom they direct their activities. Although the survey sample was directed towards policy actors ostensibly already involved in the policy process, or at least professionally attached to the issue, there was wide variation in levels of participation among these individuals , as well as in institutional context and individual characteristics. Background Literature on Political Behavior Participation in the policy process is a vital part of a functioning democracy. In an open, transparent democracy, political behavior is crucial to claims of representativeness and accountability, particularly in a plurali stic political system such as the United States. Public policy , as defined by Dye, encompasses both government action and inaction what governments choose to do as well as not to do (2002) . Participation in the public policy process can provide input for decision makers; it can bring about learning and new information perhaps even policy change and is fundamental to democratic norms (Nohrstedt, 2013; . There has been extensive work on political participation generally (Brady, Verba, & Lehman Schlozman, 1995; Hillman & Hitt, 1999; Kaase, 2009; Milbrath, 1965; Mutz, 2002; Smith, 1984; Verba & Nie, 1987; Webler, Tuler, & Krueger, 2001) . Milbrath put forth an operational definition of political participation as voting, campaign activity, community participation, and particularized contacting (1965). T here has also been considerable theorizing about both the determinants of and outcomes of particip ation . Early empirical work on political participation was dominated by voting studies as researchers explore d what mot ivates citizens to

PAGE 17

8 vote , and this continues to comprise the bulk of work on political behavior (Kaase, 2009) . Unconventional forms of political pa rticipation, such as protesting, are gaining ground in the empirical literature but nonetheless, the focus remains on convention political participation as advanced stud ies unpack the multi level and interrelated determinants of participation. The current literature regarding political behavior is limited in thre e major ways. First, knowledge surrounding political behavior narrowly focuses on elections and legislative issues. The policy process is much broader and can encompass everything from the antecedents of policy change through implementation. Second, theori zation and empirical work in political behavior tends to focus on the general public and ignores elites and other interested parties. Finally and relatedly , the concept of political behavior is often limited to participation such as voting. This is only on e aspect of political behavior. Some of the foundational ideas regarding political behavior most often participation within the policy process literature came from Kingdon, who drew upon participation as a means to illuminate the role of solution devel opment in explaining policy change (Kingdon, 1984) . Nearly ten years later , Roberts and King used grounded theory to explore how highly active individuals in the policy process might decide whether to participate, and if so, how (1991) . The work of these scholars gave some clarity about the determinants of participation and how participation might contribute to outcomes. Bishop and Davis undertook a thorough review of the extant literature surrounding political behavior in the policy process and concluded that the typologies and frameworks remain disjointed and in need of empir ical testing (2002) . They define p The authors assert that participation happens as the result of an understanding of shared power between citizens and the government . For Bishop and Davis, participation is possible only when citizens believe that their actions will

PAGE 18

9 impact government policy. From this lens, the authors divide the field of policy participation studies into approaches along a continuum of realized citize n expectations. While these approaches all have positive contributions to understanding participation in the policy process, Bishop and Davis ultimately (2002: 14) and wo uld benefit from clarity ga ined through empirical research. More i deas about political behavior with in the policy process can be found in the major theories of the policy process. Schattschneider (1960) , Lowi (1972) , and Wilson (1973) all recognized the iterative cycle between policymaking and individual political beha vior. The policy feedback literature explores how a policy can impact civic engagement (Mettler, 2002; Pierson, 1996) exogenous (Moynihan & Soss, 2014: 321) . Punctuated equilibrium theory (PET) rests upon the assumption that ind ividuals participate in the policy process in a myriad of ways that ultimately push policy making from maintenance of the status quo through incremental changes to major shifts in policy (Baumgartner & Jones, 1993; B. D. Jones & Baumgartner, 2012) . In a massive empirical effort, Baumgart ner et al. interviewed 315 individuals involved with lobbying in Washington , D.C. and analyzed hundreds of thousands of contextual documents to explore the relationship between lobbying and policy change (Baumgartner, Berry, Hojnacki, Kimball, & Leech, 2009) . While demonstrating empirically the theoretical conclusions of PET, the authors also explored how interest group arguments are shaped and the tactics that advocates employ. (1993) arg ues that policies are shaped by the political behavior of policy actors, and vice versa. The theory asserts that political

PAGE 19

10 behavior can be socially constructed within the policy process as target populations translate signals regarding their societal statu s from enac ted policies and policy makers . Fundamentally, powerful influence on public officials and shapes both the policy agenda and the actual design of tical behavior is motivated by the messages that we receive regarding target populations. theorists has lent some clarity to political behavior, additional e mpirical and theoretical knowledge would contribute to a better understanding of this mechanism within the policy process . Scholars appreciate that we need to increase empirical and theoretical awareness of how individuals participate in the policy process . Rydin and Pennington agree that theory regarding political behavior should be grounded by exploring the incentive structures of the involved individuals to ascertain how those individuals choose to participate, and what the resulting outcomes might be (2011) . Individual political behavior varies across institutional contexts, and there is a lack of empirical research that identifies how different institutional contexts affect behavior (Nohrstedt, 2013) . Zaller developed a model of public opinion formation that posited that citizens receive cues from policy elites and form opinions from which they are unlikely to deviate (1992) . (Kam, 2012) , Zaller model of public political behavior explains the role of intermittent information processing. s been highly influential in developing a research agenda on public political behavior that has produced a wealth of empirical findings and theoretical insights (see Entman, 1993; Inglehart & Welzel, 2005; Lupia & McCubbins, 1998 for some of the more notable contributions) .

PAGE 20

11 T aking into consideration the significant contributions of this body of work , the political behavior of policy actors specifically as opposed to the general public should be considered separately because policy actors operate under different conditions. For example, p olicy actors are ostensibly more experienced and may have more motivation for involvement ; however, a ll policy actors are not able to partici pate in all activities or collaborate with whomever they please (Irvin & Stansbury, 2004a) . A concerned citizen is not necessarily given the opportunity to provide testimony in Congress, regard less how much they may want to be involved. Public officials cannot participate in protests, regardless how much they may want to express their opinions. Nonprofits may simply not have the contacts n or the capacity required to develop policy networks with private industry counterparts, regardless how vital they may perceive that interaction to be in achieving their goals. There is differentiation among indi viduals as to which activities and policy networks are available , and this differentiation may be the result of contextual interaction with individual characteristics . In summary, d espite some theorizing and empirical work , there remains a lack of solid empirical evidence that explains what influences differentiation in individual political behavior among policy actors . This dissertation addresses that gap by contributing empirical research that explains how institutional context and indi vidual characteristics interact to explain the variation of political behavior within the policy process. 2 Political behavior within the policy process is crucial to our construction of political decision making. It is the amalgamation of the activities and network efforts of individuals involved in the policy process. Political behavior can provide input for d ecision makers; it can bring about learning and innovation perhaps even policy 2 All types of participation in the policy process are defined as political behavior within this dissertation. Political behavior is not limited to the activities only, as the extant literature on participation would i mply.

PAGE 21

12 change and is fundamental to democratic norms . Through an enhanced understanding of political behavior, we better understand the resulting shape and intensity of political interactions , which are critical in a well functioning democracy . Conceptual Framework This dissertation explores political behavior primarily through the lens of the Advocacy Coalition Framework (ACF). The Advocacy Coalition Framework (ACF) provides a useful lens for studying political b ehavior within policy subsystems , as the framework recognizes the role of individual behavior as a driving force within the policy process (Nohrstedt & Olofsson, 2016a) . Acco (2011) , and these actors endeavor to impact policy through coordination of behavior in an intentional manner . By exploring policy subsystems as the basic unit of research , it is possible to design research that is generalizable to other contexts as well as other policy issues. The ACF explores three main areas: advocacy coalitions, policy change, and policy oriented learning. The ACF endeavors to explain the policy process by understanding these thr ee areas. This dissertation accepts the assumptions of the ACF regarding a modified version of methodological individualism and bounded rationality and will focus on exploring individual political behavior within coalitions. Based on shared beliefs, policy actors organize into coalitions (Sabatier & Jenkins Smith, 1993) . Policy change can be brought a bout as policy actors within coalitions employ political behavior to impact the policy process (Blomquist, 2007; Jenkins Smith, Nohrstedt, Weible, & Sabatier, 2014; Real Dato, 2009; Weible, 2006) . The ACF assumes that within coalitions, policy actors employ political behavior to impact outcomes. Activities and policy networks are the

PAGE 22

13 means through which policy actors strive to influence the policy process (Heikkila, Pierce, et al., 2014) . Politica l behavior is directed and constrained by the rules and norms of participation in the policy process, which are shaped by the political context of the subsystem (Boehmke, Gailmard, & Patty, 2005; Cross, 1999; Eisenberg & LoPucki, 1999; Guiraudon, 2000; Holyoke, Brown, & Henig, 201 2; Howard, 2007; Jenkins Smith et al., 2014; Karch, 2009; Ley, 2014; Pralle, 2003) . Numerous scholars recognize that the ACF is lacking in theory regarding how policy actors make political behavior decisions in varying contexts (Elgin, 2014; Jenkins Smith et al., 2014; Nowlin, 2011; Weible, 2006; Weible & Sabatier, 2006) , but t he ACF does provide some direction about how co ntext may be influential on differentiation in political behavior . Sabatier (1998) noted that there are attributes of the available settings in which policy actors may participate that can influence behavior. For example, a more prestigious forum forces pr ofessionals to participate. Jenkins Smith explored this idea in early ACF theory, noting that which analysts promote, refute, or adjust to potentially threatening (1990: 97) . Jenkins participants. The ACF assumes that within a contested subsystem, individuals form and maintain coalitions through which they work to influence the policy process (Sabatier & Weible, 2007) . Empirical work drawing upon participation in extant ACF empirical research most often explores partici pation as a determinant in the formation of advocacy coalitions (c.f. Ingold & Varone, 2012; Nohrstedt, 2011) or as an explanation for policy change (c.f. Albright, 2011) . There is also (c.f. Webler, Tuler, & Krueger, 2001) , but

PAGE 23

14 this work focuses on the processes within participation and disregards the con textual factors that influence political behavior . A clear need remains for further empirical and theoretical work in this area to consider the attributes of varying contexts and the impact on political behavior (Henry, Ingold, Nohrstedt, & Weible, 2014a) . Against the backdrop of country variation, Henry et al. discussed the potential influence of differing contexts on policy outcomes ( ibid. ). There has been much discussion within the field regarding whether the ACF is applicable in different political systems (Henry et al., 2014a; Nohrstedt & Olofsson, 2016b, 2016a) . These discussions recognize the role of subsystem context as influential in determining political behavior and ultimately policy outcomes, and the same argument can be extended to a sectoral level. As noted by Henry et al. within the same constitutional structure, po licy subsystems display different opportunities and There is not yet consensus on the types of contextual variables that may be most influential, and this dissertation a ims to contribute to this discussion. Different institutional contexts, expressed through sectoral differences, may also vary opportunities and constraints for political behavior. Differing context s create different operating environment s for policy actors, with potentially different strategies available to influence the policy process. For example, private foundations are expressly prohibited from lobbying by the Tax Reform Act of 1969. This is an institutional constraint imposed by sectoral affilia tion within the nonprofit sector that shapes the opportunities private foundations have to influence the policy process, as recent evidence has shown to be increasingly occurring (Reckhow & Snyder, 2014) . There has not been sufficient research on how different contexts may shap e political behavior

PAGE 24

15 within subsystems (Weible, 2006) ; s upplement al theory is needed to move beyond what is currently recognized about the role of institutional context , and particularly its interaction with individual level characteristics, in the ACF literature. To supplement the ACF and generate new theory , this dissertation will draw on institutionalism , behavioral economics , policy networks, advocacy literature , conflict expansion , venue shopping , and agenda setting. Objective 1: Role of Individual Characteristics To describe how individual characteristics shape environmental policy conflicts. Understanding the policy process that underlies the development of environmental policy and the policy process, in general, requires understanding the individuals involved in environmental policymaking. According to the ACF, individuals are motivated to pa rticipate in policy activities and form coalitions to impact policy change owing to shared beliefs regarding, in this case, hydraulic fracturing. These individuals are boundedly rational and forced to call upon pre existing belief systems to process their opinions and perceptions of a policy issue. Individuals often display shared beliefs and largely stable preferences regarding their policy positions and risk assessments. Importantly, individuals are motivated to form and maintain coalitions based on share d beliefs, and it is within these coalitions that policy actors endeavor to impact policy. Objective 2 : Activities To identify patterns in the types of activities used by policy actors in the policy process and whether these patterns are influenced by ins titutional context and individual characteristics .

PAGE 25

16 One fundamental tenet of the ACF is that coalition actors will use different activities to influence policy (Weible & Sabatier, 2006) . Activities are defined as the actions taken by policy actors to influence the policy process. Examples of activities are writing an editorial, sponsoring a bill, testifying in the legislature or filing a lawsuit. Activities are the opportunities for policy actors to engage in a subsystem in a variety of ways that are created and shaped by the rules and norms of participation. There are numerous ways in which activities of competing coalitions can differ. Within the ACF, activities are the byproduct of the r esources of coalitions; the ACF assumes that coalition actors have specific resources that constrain their choice of strategy. According to Sabatier and Weible, coalition resources include: 1) legal authority, 2) public opinion, 3) information, 4) mobiliza tion, 5) financial resources, and 6) leadership (Weible, 2006) . T his objective expands on this work recognizing context as influential in determining political behavior by explor ing how activities are shaped by institutional context and individual characteristics. However, the research aims to identify many additional f actors that may enable or constrain activities. Since public policy began as a focused area of scholarship , researchers have been exploring the factors that influence participation in the policy process (Smith, 198 4) . Under the banner of conflict expansion, Real Dato undertook a detailed theoretical synergy of four major policy process frameworks regarding their respective interpretations of the influence of context, particularly institutional context, as they re late to strategic participation in the policy process and ultimately affect policy change (2009 ) . He asserted that context is a major determinant of participation decisions and called for empirical applications exploring how context impacts individual behavior within the policy process. Institutional factors, influenced by sectoral affiliation , m ay be central to the participation decisions of policy actors in a contentious policy

PAGE 26

17 subsystem. Sectoral affiliation establishes the opportunities, constraints, and incentives of individuals within those institutional settings. The institutional context i mposed on individuals in private versus public versus nonprofit sectors implies that the activities available to individuals in these differing sectors would vary. I nstitutionalism has long called for explicit attempts to contextualize analysis and rational choice institutionalism searches for generalizable explanations of decision structures (Thelen, 1999) , but few studies in the broa der field of p olitical behavior consider how context and individual characteristics might interact to influence political behavior. Sneiderman and (Sniderman & Levendusky, 2007) . They combine the views of rational choice theory and the insights of political psychology to create a model of behavioral economics. The authors recognize that both or; they mechanisms an internal one to account for choice between alternatives plus an external one to object ive adopts this same view: the political behavior alternatives on offer to an individual policy actor are determined in part by context and the choice among those alternatives is motivated by individual characteristics. Obj ective 3 : Policy Networks To ide ntify variation in policy networks and whether this variation can be attributed to differences in institutional context and individual characteristics . Individuals are constrained by time, financial resources, capacity, and most importantly, institutional rules and norms that vary by sector . For example, there are regulations and norms

PAGE 27

18 constraining the actions of nonprofit and public sector employees in regards to t heir efforts to impact the policy process. Owing to limited capacity for action, p olicy actors are forced to make strat egic decisions and target particular individuals or organizations . The policy networks that arise with in a policy subsystem may be driven by these constrained decisions. Policy networks can be viewed as the aggregation of many individual level strategic choices. Policy actors could maximize their limited capacities by targeting those organizations and individuals whom they feel would have t he highest potential impact on the policy process. The third objective explores how institutional context and individual characteristics in fluence policy networks . Several scholars recognize the importance of policy networks (Agranoff & McGuire, 2001; Hill & Lynn, 2005; Lube ll, Schneider, Scholz, & Mete, 2002; . Much of the empirical policy network research focuses on collaborative networks under the guise of governance studies. This research often recog nizes the role of constraints and competition within an advocacy space (see Lee, Lee, & Feiock, 2012) , but does not advance a full network theory of the policy process as it fails to explain the charac teristics of the components of a network the individual (Dowding, 1995) . T his dissertation moves a step beyond by exploring not only with whom a policy actor chooses to connect but also the level of importance that policy actor assigns the inter action. By giving a measure of importance, the policy actors reveal the utility that he assigns that interaction, which can yield deeper insight about strategic network choices. With the individual as the unit of analysis, as opposed to the majority of stu dies done at the organizational level, this research can explore the institutional opportunities, constraints, and incentives in combination with individual characteristics that influence strategic network choices.

PAGE 28

19 Consequently, we better understand the in fluence of institutional context and individual characteristics on a policy network . There have been several studies of individual influence on strategic policy network formation (Hadden & Jasny, 2017; Johari, Mannor, & Tsitsiklis, 2006; Watts, 2001) , but there are fewer studies that consider how the constrained capacity of policy actors influences their policy network. Leifeld and Schneider recognized the role of political opportunity structures and transaction cost analysis in individual decision ma king regrading who policy actors choose as contacts (Leifeld & Schneider, 2012) . Using the case of information exchange, the authors determined that policy actors knowingly attempt to minimize transaction costs and maximize returns within their constrained capacity. This chapter builds upon this finding by capturing the strategic choices of policy actors in network connections, as influenced by institutional context and individual characteristics. Hadden and Jasny r political behavior (2017). The authors considered how network can influence their tactical choices. Looking at transnational nongovernment organizations ( NGOs), results indicate that organizations are more likely to adopt a tactic if an organization with whom they are adjacent or directly connected has already adopted that tactic. Position in the network was found to be more influential than contextual equi valency, in that organizations were not likely to mimic similar organizations if those organizations were not directly connected, regardless how similar those organizations may be contextually. As the how interactions among actors shape (2017: 1) . In 2011 , Ingold built upon previous studies recognizing the role of individual beliefs in network formation by studying if cooperative beliefs result in collaboration

PAGE 29

20 (Ingold, 2011) . The author argues that although beliefs were shown to be instrumental in predicting network cohesion, institutional settings may constrain or enable network ties and may in fact be a better predictor. Lee et al. explored the structural features of an interorganizational network to answer questions surrounding the emergence of policy networks, the forces that shape network formation, and how networks evolve over time (2012). The authors found that organizations seek to build stable and sustainable relationships, which require more than simple exchange transactions. There appears to be some level of strategic awareness among organizations of the importance of participating in a dense network with credible ties. Theories of behavioral economics adopt tools of economic analysis to exp lore , for example, political issues. Behavioral economics assume s that individuals act strategically and in their self interest to maximize their utility (Downs, 1957; B. D. Jones, 2001; Shepsle, 2008) . Thes e assumptions are not limited to economic choices; this applies political behavior in choice of network targets as well. Theoretical economic models of individual strategic choice in network formation have been developed (Bloch & Jackson, 2006; Galeotti, Goyal, & Kamphorst, 2006; Goyal & Vega Redondo, 2005; Jackson & Watts, 2002; Jackson & Wolinsky, 1996) but empirical applications are lacking. To better understand micro level political behavior within the policy process, this objective focuses on another aspect of political behavior: policy networks. The unit of observation is again the individual policy actor, but this objective will explore a policy network as individuals strategically choose which organizations to interact with in order to achieve their policy goals. Many studies in empirical ACF work use networks to identify coalitions (Fischer, Ingold, Sciarini, & Varone, 2012; Lienert, Schnetzer, & Ingold, 2013) , often recognizing the role of shared beliefs as the cohesive mechanism for network formation (Ingold, 2011; Ma tti &

PAGE 30

21 Sandström, 2013; Weible, 2005; Zafonte & Sabatier, 1998) . As Ingold notes, coalitions are often reflected in relations among actors The research proposed here does not dispute that shared beliefs can encourage the formation of policy coalitions as networks but rather aims to bring more theoretical clarity to that assumption by test ing whether tie formation might also be motivated by context . There is substantial empirical evidence that belief homophily is a driver of network formation (Jenkins Smith et al., 2014) ; this research seeks to explore other contextual variab les, such as institutional conditions, that may influence network choices as well. This objective is built upon previous network studies in the ACF to expand knowledge ontext and individual characteristics may influence network structure. The goal of this objective is to identify and explore political behavior as networks reflecting who an individual policy actor targets to achieve his or her policy goals. These are related to but may also be distinct from coalitions. This study identifies coalitions as a basis for t he network analysis, then explore s further differentiation into policy networks based on sectoral affiliation . This parsing alone yield s some explanation regarding how, for example, the policy network of nonprofit affiliated individuals is structurally dif ferent than the policy network of private sector individuals. The policy networks in this study are not intended to represent or define coalitions 3 , but rather will be differentiated by institutional conditions and individual characteristics. 4 In this stud y, p olicy 3 There are several reasons why this particular network analysis does not intend to define or represent coalitions. First, the network is unidirectional and not symmetrical, in that the ties are not necessarily reciprocal. Second, the network is als o bipartite, or two mode, composed of two different types of vertices, in this case individual organization. Finally and relatedly, the network is incomplete, as is common in social network analysis. 4 It should be acknowledged that s ome network studies h ave defined and i dentified coalitions using the type of network described proposed here . There is a sizeable literature within the ACF that debates how coalitions should be defined: by beliefs, or by network connections, or by shared venues, among others. Whether network connections,

PAGE 31

22 actors holding opposing positions may find themselves in the same policy network when conditioned on institutional context such as sector instead of on stance . The ACF is a framework that enables multiple theoretical arguments (Jenkins Smith et al., 2014) , and while the goals of ACF assumptions . There is overlap in this question compared to what is traditionally asked in ACF coalition studies. Researc h Design Th is dissertation is multi method case approach that aims to inform about a larger group of cases: high conflict policy issues in general (Gerring, 2007a) . The goal is to inform the larger from the smaller, generalizing the explanations that result from this research to the broader population of high conflict policy issues. While the ultimate aim of the research is explanatory, case study description, this research could also be characterized as a thick by Blatter and Blume (200 8), as opposed to Gerring . According to Gerrin g invariably face a choice about knowing more about less, Blatter and Blume argue instead that thickness should not immediately dictate a tradeoff in generalizability. 5 Thickness should instead be determined by a multiplicity of observations within the case as well as lanation of the relationships discovered. It is in these cases that applicability of our findings to broader research paradigms can be maximized , and within which this particular case study aims to contribute. particularly unidirectional network connections, are sufficiently cohesive in coalition formation, remains an open empirical question (see Henry, Ingold, Nohrstedt, & Weible, 2014) but not one for this dissertation. 5 Blatter and Blume (2008) explore the covariance form of case studies , in which variables centered testing domi nates the analysis. Objectives 2 and 3 of this dissertation conform to this form. As the authors explain,

PAGE 32

23 Case Selection With its a multi method case s tructure and aim to uncover patterned influences of context, the dissertation is novel in its approach to first use data from the national level without regard for contextual variation. This sets up the initial understanding of the role of individual chara cteristics. The study then moves into the specific case of Colorado, which is a typical case and by definition is representative (ibid.). Identifying a case as representative requires some knowledge about the characteristics of the broader populat ion. A representative case of high conflict policy issues should demonstrate continuity in the salience of the issue, maturity of the subsystem, and variation in political ideology and theoretically driven e xpectations on factors that might dampen or bolster conflict. Hydraulic fracturing has been in use in Colorado for decades, and while oil and gas revenues are sizeable enough to make the issue highly salient to the general population, the state is not sol ely dependent on revenues from production. In contrast, a state such as Texas would be a poor choice as a typical case as it is much more dependent, relatively, on oil and gas production. Texas would be better identified as an extreme case. There are areas within Colorado that are more dependent on oil and gas revenues than other areas but this is important to the expectation of issue salience. It is reasonable to expect that political ideology may contribute to levels of conflict on policy issues, and henc e variation in political ideology is important for a typical case. Political leanings in Colorado have been fairly evenly split for the last century, with nearly the same number of Democrats and Republicans elected as Governor, for example (Bump, 2015) . As with much of the rest of the country, there is an ideological divide between urban and rural areas, with urban areas and some mountainous regions pulling the

PAGE 33

24 slightly left, but the state has been a swing state in the last two presidential elections and is expected to continue to be relatively open ideologically. This research uses an exploratory and explanatory approach to understanding the case of hydraulic fracturing in Colorado . The aim of the research is to uncover and explain the political behavior expressed by policy actors in different contexts. A non experimental case study , such as this , is designed to explore the limits of exis ting theory and possibly build new theory (Van de Ven, 2007) . Van de Ven notes that non experimental case studies may be useful in construct validity, drawing upon theoretical inference to develop better constructs within our the ories. (Singleton & Straits, 2010: 141) . This case meets the preconditions of the covariance approach, as first detailed by Gerring (2007a) and further explicated by Blatter and Blume (2008) . The approach demands control of exogenous variables through case and indicator selection and theoretically driven hypotheses regarding the causal direction of relationships. Case studies applying the covariation approach observe, over time and/or space, variance in the independent an d dependent variables of interest. The conclusion is therefore that X has a causal effect on Y . C ovariation case studies are focused on theory, not on explicit aims of explanation of the case itself (Blatter & Blume, 2008) . The covariance case study seeks not to achieve consistency within empirical r esults but rather within more abstract concepts. This is crucial to claims of generalizability and theory testing. Therefore, the purposive choice of the case of hydraulic fracturing in Colorado aims not to explain the exploration of unconventional oil and gas but rather to use the insights gained through this case setting to explain political behavior within high con flict policy issues in general.

PAGE 34

25 Methods of Data Collection The analysis draws upon a set of surveys and interviews completed at the national level in 2016 and in Colorado in 2015 and 2017. The in depth interviews associated with the surveys served three s eparate purposes. First, the interviews informed the development of the survey. Second, the interviews helped identify additional policy actors and activities not picked up in the mapping process as a form of modified snowball sampling. Snowball sampling w as developed as a way to uncover the structure of social networks for which the population is not known a priori (Douglas D Heckathorn, 2 011) . Modified snowball sampling attempts to correct for the issue of hidden populations inherent in any social network. Previous experience within the policy issue generated an initial set of interviewees, after which modified snowball sampling was used to provide names of any other individuals that key stakeholders felt were important to interview as well. Finally, the int erviews provide richer context for the quantitative data gathered through the survey. A mixed methods approach in case study research design can be more impactful in this simultaneously exploratory and explanatory study (Riccucci, 2010; Van de Ven, 2007) . Sampling: Policy Actor Identification The first step in any ACF application is to identify the policy actors involved in the subsystem. One of the basic assumptions of traditional ACF theory is th at it adopts a viewpoint of a decade or more. Weible and Sabatier have noted that the ACF can consequently be difficult to apply as it is time consuming and challenging to gather details about the actors involved over such a long timeframe (2006) . ACF scholars have encouraged researchers to de velop creative, more accessible, and possibly quicker methods to map policy subsystems (Jenkins Smith et al., 2014; Weible & Sabatier, 2006) .

PAGE 35

26 This research first map ped the policy subsystems to uncover the policy actors involved in the subsystem through online identification and reporting of involvement in the policy issue. This method utilizes advancements in techn ology and worldwide communication since the inception of the ACF. However, there are limitations to the method, although many of the limitations apply to traditional mapping methods within the ACF literature as well. First, it can be difficult to identify hidden or silent policy actors in the subsystem without expert knowledge of the subject area. Second, this method of subsystem mapping relies exclusively on publicly available reports and accounts of actor activity. There are other ways in which policy act ors participate in the subsystem that are not publicly available. This limitation will be addressed in later steps of the data collection process. Third, the researcher must have at least a basic understanding of the context and policy processes in the sub system in order to uncover the activities of the actors. Social science research has begun to recognize the usefulness of the internet as a tool (Singleton & Straits, 2010) . The subsystem mapping and policy actor identification started with an online search of t he policy issue to include legislative proceedings, court proceedings, media and any other venues in which policy activity is possible. The search terms were /United States AND [hydraulic fracturing OR fracking OR unconventional oil OR unconventio within the policy issue were identified as potential policy actors. Media including but not limited to blogs, reports, press releases, articles, and conference pre sentatio ns was also searched to identify active individuals . Any publicly available document that records evidence of participation was used. Finally, individuals and organizations now identified were cross searched with the original search terms to explore othe r activity that was not initially found.

PAGE 36

27 The sample for the survey was purposive. The survey sample is not intended to be a reflection of public opinion; it is intended to be a survey of the perceptions and attitudes of individuals who are alrea dy engage d in the policy debate . Survey questions were developed after in depth interviews with key policy actors in each state. The survey was reviewed in its entirety by an advisory board of influential and experienced individuals who are highly engaged in the topic of hydraulic fracturing. See Table 1 for a summary of the data collection results of the two waves o f the surveys. The unit of analysis is the individual. Some predictors are measured at the political behavior decisions and do not therefore necessitate a need for cluste r ing. Table 1: Data Sources Chapter Geographic Scope Source (n) Date Gathered Level of Observation Contents Analysis Chapter 1 National Survey (133) Interviews (11) Spring 2016 Individual Activities Network Context Beliefs Stance Threats Controls Statistical description Basic correlations Chapters 2 and 3 Colorado Survey (213) Interviews (9) Spring 2015 Individual Activities Network Context Beliefs Stance Threats Controls Factor a nalysis Multinomial logistic r egression Social network a nalysis ERGM Qualitative t riangulation Chapters 2 and 3 Colorado Survey (186) Interviews (29) Spring 2017 Individual Activities Network Context Beliefs Stance Threats Controls Factor analysis Multinomial logistic regression Social network analysis ERGM Qualitative triangulation

PAGE 37

28 Key Variable Operationalization T his research uses the bare essentials of a concept, sufficient to differentiate it extensionally without excluding any (Gerring, 2012: 135) . Because the aim of this research is to explore all possible combinations of political behavior used by policy actors to influence the policy process, a minimal definition of variables is preferred. In addition, many of these concepts are nebulous and adopting a parsimonious definition with widespread understanding can be a n effective strategy (Gerring, 2012) . Appendix A details constructs and operationalization for all key variables . Activities will be oper ationalized as an action that an individual takes to participate in the policy process with the intent to achieve their personal or professional goals in related t o oil and gas development that u ses hydraulic fracturing (r efer to question 17 in Appendix B : Full Survey ) . Each respondent was given a choice of eight different activities and asked to rank the effectiveness of those activit ies on a Likert type scale from Not Engaged to Very Effective. The activities were: Brokering agreements between parties, Countering arguments made by people you disagree with, Mobilizing the public, Collaborating with people you disagree with, Coordinatin g political activities with allies, Providing information to government officials, Providing information to the news media, and Sharing your opinion with government officials. Previous versions of the survey asked respondents to identify how frequently the y participate in a similar battery of activity perception of the effectiveness of different activities, instead of limiting the information gathered to simply a measure of frequency. The s pectrum of activities was vetted through interviews, review ed by an advisory council of issue experts , and tested in previous versions of the survey to

PAGE 38

29 ensure that it represented a reasonable majority of possible ways in which a policy actor might engage i n the policy process. Policy networks are the networks of organizations and individuals that are accessed by policy actors to achieve their personal or professional goals related to oil and gas development that uses hydraulic fracturing. Respondents were given a battery of fifteen organizations Federal government, State government, County government, City government, Oil and gas industry, Oil and gas industry, Environmental or conservation groups, Real estate developers or home builders, Agricultural organization or farmers, Organized citizen groups, Churches or other religious organizations, Universities or colleges, Consulting firms or think tanks, Informal personal networks, and News media and asked to rank importance of interaction on a Likert type scale from not at all important to very important, with an option for not applicable (refer to question 9 in Appendix B : Full Survey) . As with the activities question, this battery of choices was tested in earlier versions of the survey an d reviewed by an advisory council . Finally, p olitical behavior is operationalized as the amalgamation of activities and policy networks . Analysis The result of the data collection is a collection of rich qualitative and quantitative data that can be used to expand theory on the nature of political behavior in different contexts. The study illuminate s the types of behaviors that could emerge and provide information about how individual s choose one activity or network collaborator over another . The simultaneously descriptive and explanatory nature of the research question lends itself to rich description of the results and may also lead to the creation of new typologies of political beha vior . Nuances of the interaction between the organizational context and individual characteristics may be used to illuminate the nature of political behavior as well as the context variables themselves. Theory

PAGE 39

30 expansion can be achieved by using the analysi s to revisit the original propositions and develop hypotheses of a more explanatory quality related to what drives political behavior decisions . In the first objective and C hapter 1 , statistical description and basic correlation is used to describe the role of individual characteristics in shaping environmental policy debates . The research explores the beliefs and policy preferences of individuals involved at the national level in the policies shaping the extraction of unconventional oil and gas in the US; it focuses on national level policy actors as these individuals in comparison to individuals active at only the state or local level should be less susceptible to local level variance. For the second objective and Chapter 2 , factor analysis is used to determine if there are patterns among activity choice by policy actors. D eterminants of those activity patterns , hypothesized to be stance on the issue, perceptions of viability, and perceptions of threat, are sub sequently tested through multinomial logistic regression. Gender, education, professional experience, stance, and political ideology serve as controls. For the final objective and Chapter 3 , social network analysis is used to first describe the policy net works that have developed around hydraulic fracturing in Colorado. The whole network is then differentiated by sectoral affiliation (public v. private v. nonprofit) of the respondent and network structure is regressed against expectations regarding strategic network choices using inferential ne twork analysis through an exponential random graph model s (ERGM). The ERGM results of each sectoral network are compared to explain variation in network structure across sectors.

PAGE 40

31 Limit ations Case study research inherently faces tradeoff of external validity versus representativeness. Generalizability can be limited owing to the small number of cases under consideration (Gerring, 2007b) . This study attempts to correct for this limitation by designing research at the subsystem level and observing within the subsystem, which may lend itself to application in other contexts. In addition , the purposive choice of a representative case enhances the potential for ge neralizability. The findings may lend themselves to empirical generalizability to other high conflict environmental policy issues, but there is potential that theoretical generalizability could be limited to other cases in which institutional context is de termined by sector. There is a risk of collinearity among the key political behavior s . However, since this study does not aim for precise causal explanation of the empirical setting but rather general theoretical observations , collinearity can be managed (Gerr ing, 2012) . One limitation of many ACF applications is the inability to locate hidden populations. There are two aspects to hidden populations. First, the study might be limited in its ability to locate or access hidden populations of policy actors who do not appear in the sampling frame. As an attempt to address this issue, identification of potential hidden populations was emphasized in the modified snowball sampling in the interviews. Relatedly, there is potential for selection bias in the sample. As the sample is intended to identify those individuals who are active in the policy process, there may be an overrepresentation of frequency of participation or a structural tendency to survey individuals whose cognitive characteristics orient them towards p articipation. The second aspect of hidden populations, however, may be more of a contribution than a (McConnell,

PAGE 41

32 2017: 1) . These policy actors may be visible but their actions may not, which is part of the popu lation id entified in this study. T his research project aim s to explain why policy actors chose the strategies they do. There is a story behind why some policy actors chose to remain hidden, which is a strategy itself and may be the result of specific polit ical, subsystem or individual resource context. Contributions While past research has explored individual components of political behavior, this dissertation offers a more holistic picture of how an individual behaves within the policy process. Research e xploring how political behaviors vary in different contexts and across time would bolster the explanatory capacity of the ACF. Expanding theory on political behavior contributes to the broader ACF framework and the overall policy process literature. With m ore knowledge and empirical evidence regarding the causal processes underlying political behavior, it is possible to refine theory to address contextual variations. The research contributes to ACF empirical work exploring and explaining how political behav ior may vary based on the circumstances under which a policy actor finds himself. Research exploring how political behaviors vary in different contexts would bolster the explanatory capacity of policy process theories. This research also has implications f or understanding policy change. In theory, policy likely to prevail (Holyoke et al., 2012; Schattschneider, 1960) . To understand influence within the policy process , it is important to understand the behaviors individuals access to create that change and whether there are systematic barriers that shape political behavior, which in turn impacts the quality of politics. Owing to institutional conditions, particular gro ups or individuals may be disadvantaged within the policy process. This dissertation can contribute to

PAGE 42

33 understanding whether disparities in political behavior are the result of structural inequalities deriving from institutional context or rather due to in dividual characteristics. The former is less encouraging for a well functioning democracy. Methodologically, the research proposed will utilize a variety of data analysis tools to explain the influence of context on political behavior. Many of these statis tical approaches, such as inferential network analysis, are not currently widely used within the ACF. There are very few applications within policy process studies that use longitudinal data to employ inferential network analysis; this dissertation is an i nfrequent example of longitudinal empirical work on political behavior. does to impact the policy process, but with whom the policy actor interacts and the importance o f interaction to better understand the governance landscape of this high contested policy issue. The first objective describes the role of individual characteristics in shaping highly contentious policy issues and sets the stage for the following two objec tives. Objective 2 produces knowledge regarding patterns of activity choice, as differentiated and determined by institutional conditions and individual characteristics. Objective 3 draws upon a novel approach to network studies and considers how instituti onal constraints and individuals characteristics strategically shape policy networks and influence the efficiency of the resulting networks. The dissertation as a whole develops nuances to the interaction between institutional and individual level charact eristics, and ultimately, informs theory and practice regarding the consequences of contentious policy exchanges. In terms of practice, this research could be used by public sector individuals, nonprofit organizations, and private enterprise to better unde rstand how their specific context may enable more effective participation in the policy pr ocess.

PAGE 43

34 CHAPTER II PERCEPTIONS OF CONTENTIOUSNESS: HOW INDIVIDUAL TRAITS SHAPE ENVIRONMENTAL POLICY CONFLICTS 6 The extraction of unconventional oil and gas using hydr aulic fracturing is a complex environmental issue that has drawn contention from all branches of government in the United States (US) Congress, the White House, and numerous federal agencies as well as state government and local municipalities. The Oba ma administration fought to regulate hydraulic fracturing on federal lands, an important step for national regulation. However, while many shallow and did not go f ar enough (Harder & Gilbert, 2015) fracturing regulations were quickly reversed by the Trump administration (Tabuchi & Lipton, 2017) . The Environmental Protection Agency (EPA) spent more than six years on a landmark study exploring the water rela released in 2015, it was met with contempt from groups both opposed to and in support of exposed cit izens to potential danger, while the latter were disappointed in the lack of an enforceable directive eliminating future questions related to the environmental impacts of oil and gas extraction (DiChristopher, 2016) . By December 2 016, the EPA reversed course somewhat and highlighted the sometimes tenable but nonetheless real connection between oil and gas operations using hydraulic fracturing and contamination of drinking water (EPA 2016) . While oil and gas production utilizing hydraulic fractu ring may have tapered off from its apex in 2015, the 6 Portions of this chapter were previously published in E nvironmental Policy and the Pursuit of Sustainability (2018) London, United Kingdom: Routledge, and are included with the permission of the copyright holder.

PAGE 44

35 debate surrounding the extraction of unconventional oil and gas remains critical for environmental and energy policy (Heikkila, Weible, & Olofsson, 2017) . The extraction of unconventional oil and gas using hydraulic fracturing hits on many environmental issues including water supply and quality, ai r quality, emissions, public nuisance, energy security, economic impacts, energy costs, and public health matters. The issues involved are complex and require involvement from all levels of government, and consequently, environmental policies in pursuit of sustainability are difficult to form. Understanding the policy process that underlies the development of environmental policy and the policy process, in general, requires understanding the individuals involved in environmental policymaking. The objective of this research is to understand the beliefs and policy preferences of individuals involved at the national level in the policies shaping the extraction of unconventional oil and gas in the US, including individuals involved in policymaking, citizens impa cted by hydraulic fracturing, individuals who work in hydraulic fracturing professionally, government regulators, environmental groups, interested media, and more. The research focuses on national level policy actors as these individuals in comparison to individuals active at only the state or local level should be less susceptible to local level variance. The purpose of this research is to understand how the individual level beliefs and policy preferences of actors in environmental decision making infl uence their views while participating in the policy process. By exploring the beliefs and preferences of individuals involved in a high conflict policy issue, and how those belief structures shape perceptions, we can better understand the development of en vironmental policy conflicts. After a discussion of the case setting, the chapter first examines the characteristics of national level policy actors involved in making policies regarding hydraulic fracturing and then

PAGE 45

36 explores how differences in characteris tics among policy actors are associated with varying perceptions of issue contentiousness. Case Setting: Hydraulic Fracturing In hydraulic fracturing, fluids and other chemicals are injected into wells at high volumes and pressure to fracture the rock and (Rahm, 2011) . The technique of hydraulic fracturing is not new; its emergence dates as early as the 1940s (Hubbert & Willis, 1957; Phillips, 1972) . The use of hydraulic fracturing increased dramatically around 2005 as advancements in horizontal drilling technologies opened up ways for the oil and gas industry to access shale oil and gas reserves that were previously not economically viable (Heikkila, Pierce, et al., 2014) . As the usage of the technique rapidly expa nded, so too did the debate. Moratoriums and bans have been discussed in several US states, from New York and Pennsylvania to Oregon. Bans have been passed in three states to date New York, Vermont, and Maryland, as recently as March 2017 with the ban in New York in 2014 garnering significant national attention. Moratoriums and local level bans have been contested in numerous highly visible cases in several state supreme courts. The hydraulic fracturing policy issue is not only important for environmen tal and energy policy conflicts, but it is also critical to understanding the behavior of individuals involved in high conflict policy issues. Hydraulic fracturing is marked by complex policy issues, interrelated and often inseparable motivations of policy actors, relevancy at multiple levels of government and regulatory agencies, and broad importance to the general public as an issue that is directly related to the economy and the daily lives of many citizens. The results of this study can help in drawing more general conclusions about high conflict environmental policy issues. It is critical for policymakers and citizens alike to understand high conflict environmental policy issues

PAGE 46

37 because there are implications for the quality of politics that emerge in t hese situations and related democratic and environmental justice concerns. In high conflict policy issues, there may be a tendency towards polarization and stalemate, which is not productive for effective or quality policy. In terms of democratic concerns, we should study environmental policy conflicts because we learn more about who is participating and how, if there are barriers to participation, and whether there is an unjust distribution of benefits or unequal burden of risk. Issues related to hydraulic fracturing are fraught with concerns regarding environmental justice. Very few individuals are directly impacted by hydraulic fracturing operations, yet many, many more individuals experience the benefits and have the power and/or access to impact policy. While this research does not endeavor to address all questions related to hydraulic fracturing policy, it does contribute to understanding how individual beliefs and policy perceptions shape an environmental policy conflict. The theoretical basis for the research is the Advocacy Coalition Framework (ACF). The ACF is a useful lens through which to view the development of high conflict policy issues. According to the ACF, individuals are motivated to participate in policy activities and form coalitions to i mpact policy change owing to shared beliefs regarding, in this case, hydraulic fracturing. These individuals are boundedly rational and forced to call upon pre existing belief systems to process their opinions and perceptions of a policy issue. We can expe ct, therefore, that individuals will display shared beliefs and largely stable preferences regarding their policy positions and risk assessments. Importantly, individuals are motivated to form and maintain coalitions based on shared beliefs, and it is with in these coalitions that policy actors endeavor to impact policy. This can become problematic for effective policy making if coalitions become silos and perceptions of high conflict issues become both polarized and stagnant. This research

PAGE 47

38 aims to interroga te the structures and perceptions of policy coalitions surrounding hydraulic fracturing in the US to explore how these coalitions are impacting the policy process. Data Collection Data collection started with a series of eleven in depth interviews during autumn 2015 with policy experts in unconventional oil and gas extraction. The policy experts were from a range of organizations, including professional industry association representatives, federal agencies, nonprofit environmental groups , advocacy groups, and government representatives. There was no geographic target for these interviews, but all interviewees were active in policy at the national level, as opposed to the state or local level, in matters concerning the extraction of oil an d gas using hydraulic fracturing. The in depth interviews were used to tailor the survey to reflect the shape of the issue at the national level, and the survey was then vetted with an advisory committee of policy experts and updated based on their recomme ndations. These interviews also contributed to development of the sampling frame. A survey of policy actors involved at the national level on issues related to the extraction of unconventional oil and gas using hydraulic fracturing was administered in the spring of 2016 (see Appendix B for the full survey) . A total of 468 policy actors were identified as actively involved in or knowledgeable about unconventional oil and gas development in the US. Individuals were included without regard for organizational a ffiliation or profession. Rules for inclusion required that the individual had to demonstrate activity within the policy issue since 2008 and had to be currently active at the national level (as opposed to the state or local level). Similar sampling proced ures have been used in past research at the state and local levels (Heikkila et al., 2017; Weible, Olofsson, Costie, Katz, & Heikkila, 2016) . The issue of hidden populations is a limitation of any samp ling procedure in which the population is not known

PAGE 48

39 apriori, as is common in social science research (D. D. Heckathorn, 1997) . Because there is no comprehensive register of policy actors, approximation methods must be used to identify the population of individuals that may be active within the issue and to review the composition of the final sample to explore possibilities of systemic errors in the creation of the survey sample. Individuals were first identified through the in depth interviews and then used as seeds for online discovery of activity (Nohrstedt & Olofsson, 2016b) . For online discovery of activity, the internet is searched for any and all forms of ac tivity as potential for inclusion: testifying at a public hearing; comments on a public forum; employment or involvement with private, public, or nonprofit organizations active within the policy space; social media; traditional media, and many more. For ex ample, if an individual with a federal agency mentions during the interview that she worked with a research group at a university, the website of the research group is found and then reviewed to potentially include any and all individuals within that group that are active within the policy issue of hydraulic fracturing at the national level. If perhaps the interviewee was connected to the university research group through a co worker in another department, then that co worker is identified for inclusion in are all reviewed for potential inclusion as well. The survey was administered via email using the Qualtrics platform. Of the 468 individuals who were administered the survey, 133 responded for a response ra te of 28%, which is considered acceptable for online social science survey research (Nulty, 2008; Topp & Pawloski, 2002) . The design of the survey was theoretically based on the ACF and drew on similar surveys administered in the same issue area (see s ummaries in Heikkila et al. 2013; Heikkila & Weible 2015)

PAGE 49

40 tradition. The survey was developed under the framework to uncover the beliefs, positi ons, and perceptions of policy actors involved in the issues surrounding hydraulic fracturing. Results There does not appear to be systematic bias in the organizational affiliation or position of respondents, limiting the potential for sampling bias. In c omparison to the full sample, there was an equal spread of organizational affiliations among respondents. In terms of stance, it is nearly evenly distributed; respondents with a stance against hydraulic fracturing report that they would like to see hydraul ic fracturing either stopped (16.8%) or limited (34.5%) in comparison with respondents with a stance in support of hydraulic fracturing that prefer production continue at the current rate (25.7%), expand moderately (14.6%), or expand extensively (8.4%). Th e average political views, and describes his priority level regarding dealing with issues related to hydraulic fracturing in either a professional or pe rsonal c . The average respondent has relatively high levels of experience in reading about issues related to hydraulic fracturing and slightly less experience researching and analyzing topics involved in hydraulic fracturing, with comparatively low levels of experie nce in planning, working, or managing oil and gas operations or living within visible proximity of oil and gas operations. Combined with the relatively high level of education within the sample, the variation in types of experience with issues surrounding hydraulic fracturing may indicate that this sample is more professionalized in their experience rather than experiencing the issue personally. Finally, the average respondent holds a position that oil and gas development that uses hydraulic fracturing shou ld be limited somewhat or continued at the current rate, in comparison to other

PAGE 50

41 respondents who take more extreme positions that hydraulic fracturing should either be stopped altogether or expanded moderately or extensively. There are some sizeable variat ions within these averages, however. Regarding education, the sample mode is for a Masters or professional degree but that only represents one third of the nearly h alf of all respondents reported that dealing with hydraulic fracturing issues is of high importance to their professional or personal goals, nearly one third reported that it is only of low to moderate importance. As noted earlier, the vast majority of the sample holds either liberal or moderate political views, with very few respondents reporting conservative or extremely liberal views. The modal category of organizational affiliation of the respondent was Environmental or Conservation Groups, representing around 38% of the sample. This was followed by Consulting Firms or Think Tanks (15%), Federal Government (11%), Oil and Gas Industry (11%), Oil and Gas Professional Associations (7%), University Research (7%), Organized Citizen Groups (4%), State Governme nt (2%), and Other (5%). Respondent Belief Structures We can now begin to explore the belief structures of these policy actors and associate those with our variable of interest: individual perceptions of the conflict, measured as a arding the level of contentiousness of the policy issue of hydraulic fracturing. Respondents were asked to rate the level of political contention regarding oil and gas development using hydraulic fracturing, in comparison with other political issues in the United States. The Likert type scale ran from 1 to 5 with the following response options: Far less contentious (1), Less contentious (2), Just as contentious (3), More contentious (4), Far more contentious (5). Table 4.2 summarizes the results.

PAGE 51

42 Table 2. Perception of Level of Issue Contentiousness Frequency Percent Cumulative Percent Far less contentious 3 2.80% 2.80% Less contentious 16 14.95% 17.76% Just as contentious 42 39.25% 57.01% More contentious 36 33.64% 90.65% Far more contentious 10 9.35% 100% TOTAL 107 100.00% 100.00% sample. The mean response for this question was 3.32 with a standard deviation of 0.94, indicating that generally, respondents find hydraulic fracturing to be at least as co ntentious if not more contentious than other political issue in the United States. Only about 18% of respondents felt it was less contentious or far less contentious. The standard deviation is relatively high in comparison to the mean, which indicates that there is fairly wide variation in the answers. However, that variation is concentrated within the group of respondents who feel the issue is at least as contentious or worse, which makes the distribution of responses largely irrelevant for our purposes. F or ease of interpretation, responses on perception of level of issue contentiousness were collapsed into two categories: Less Contentious (n=61) and More Contentious (n=46). Figure 4.1 presents the individual traits discussed in the previous section, diffe rentiated by perception of contentiousness.

PAGE 52

43 F igure 1: Individual Traits by Contentiousness It is encouraging that there are not strong patterns within the sample regarding stance, contentiousness. Strong patterns in these basic individual traits such as educ ation and stance might indicate sample bias or that the particular policy issue of hydraulic fracturing is defined by its complexity or along partisan lines. There appear to be some differences regarding experience n of the level of contentiousness of hydraulic fracturing, but it is important to note that none of the differences between and among groups were found to be statistically significant through two 2 test). 0 5 10 15 20 25 30 35 Less Contentious More Contentious Stance Pro Stance Anti Stance 0 5 10 15 20 25 30 Less Contentious More Contentious Political Position Liberal Moderate Conservative 0 5 10 15 20 25 Less Contentious More Contentious Education B.A. Master's Degree Law Degree Ph.D. or M.D. 0 5 10 15 20 25 30 Less Contentious More Contentious Experience Less Experience Moderate Experience High Experience

PAGE 53

44 In absol ute terms, however, one of the more noticeable patterns is in professional and personal experience, which has also been collapsed into three categories for ease of interpretation. Individuals with less experience have a tendency to perceive hydraulic fract uring as less contentious, in comparison with individuals with moderate or high experience, who have a tendency towards a perception that the issue is more contentious. Caution is warranted in drawing generalizations regarding the association between exper ience and contentiousness, as this measure is rather blunt. It collapses several different types of experience, such as researching or analyzing the impacts of hydraulic fracturing, into one category with other types of experience, such as owning mineral r ights or living in visual proximity of oil or gas operations. To check for construct validity of the collapsed measure, all individual types of experience were tested individually against contentiousness in two ways measures of association Chi 2 square d effect size for regulating experience indicates that this type of experience accounts for nearly 4% of the variance in perception of issue contentiousness, which appears relatively small but is still statistically significant. Even more pronounced is the omega squared effect size for participating in political activities, which indicates that this type of experience accounts for more than 10% of the variance in perception of issue contentiousness. It appears that experience with particular types of activi We can now begin to explore the belief structures of individual respondents to understand how cognitive characteristics, such as perceived threat, may be associated with per ceptions of issue contentiousness. Recent work on policy conflicts has shown that individual

PAGE 54

45 traits like education and stance are not as consistently influential upon perceptions of contentiousness as are cognitive characteristics such as threat perception or dissenting relationships (Heikkila & Weible, 2017; Weible & Heik kila, 2017) . Figure 4.2 illustrates the associations between four cognitive characteristics and perception of level of issue contentiousness. Note: Statistical Significance: ***p<0.01, **p<0.05, *p<0.10 Figure 2: Cognitive Characteristics by Contentiousness All of the associations in Figure 4.2 were found to be statistically significant, indicating that there are non random relationships between these cognitive characteristics and perception of 0 10 20 30 40 Less Contentious More Contentious Personal Threat** Very Little Quite A Bit 0 5 10 15 20 25 30 35 40 Less Contentious More Contentious Theaten United States* Very Little Quite A Bit 0 10 20 30 40 Less Contentious More Contentious Relationships* Not Collegial Somewhat Collegial Very Collegial Completely Collegial 0 10 20 30 40 Less Contentious More Contentious Issue Priority*** Low Priority High Priority

PAGE 55

46 issue contentiousness. Respond ents were asked to gauge how the actions of those you disagree with on oil and gas development that uses hydraulic fracturing threatens both themselves and the Unites States. Response options were on a Likert type five g States at 3.40; however, when associated with issue contentiousness, differentiation is more pronounced for personal threat. Policy actors appear to associate thre at and issue contentiousness: the less an individual feels threatened personally by the actions of those with whom they disagree, the less they feel that the issue of hydraulic fracturing is contentious, and vice versa. The same pattern holds for threat ag tolerance is directly linked to their perception of issue contentiousness. The omega squared score for personal threat is 0.081, indicating that perception of personal threat accounts for 8.1% of the variance in perceptions of issue contentiousness, while perception of threat towards the United States, although statistically significant, accounts for a much lower 1.8%. Respondents were also asked to describe the collegiality of their relationships with people they disagree with on the issue of hydraulic fracturing. Responses options were again on a Likert type four collegiality of relationships with those with whom a respondent disagrees was 2.23, around opposite sides of an issue, a low mean is a valid result. The association between relationships with those with whom you disagree and contentiousness is slightly convoluted but it is ns on the issue of hydraulic fracturing. This is an intuitive result that if a

PAGE 56

47 policy actor feels that their relationships are extremely collegial, even among those with whom they disagree, then it seems logical that the same policy actor would view the is sue of hydraulic fracturing as less contentious. The patterns among the moderate responses on relationships do not display an obvious trend. The collegiality of relationships among policy actors who disagree accounts for nearly 4% of the variance in percep tion of issue contentiousness. In regard to issue priority, there are strong and statistically significant patterns associated with perceptions of issue contentiousness, but in ways that do not conform to patterns observed in previous respondent cognitive priority is it for you professionally or personally to deal with political and policy issues related to educated and more professionally than personally experienced on the issue of hydraulic f racturing, so this question may be picking up on that description. Nonetheless, there is a strong association that if an individual feels the policy issue of hydraulic fracturing is a high priority, they are also much more likely to perceive the issue as c ontentious, and the reverse holds for respondents who feel that the issue is of low priority. Finally, respondents were asked to consider a measure of a policy outcome: protection of the environment and public health. Specifically, respondents were asked to gauge if protection of the environment and public health had become worse, stayed the same, or become better in the past two years. We might expect that those who believe that environmental and public health outcomes had deteriorated in the past two yea rs would also perceive the issue to be more contentious. However, that expectation is not confirmed (see Figure 4.3). Although the

PAGE 57

48 association between environment and public health outcomes and issue contentiousness is strongly statistically significant, d iscernable patterns are not readily evident. Figure 3: Environmental and Public Health Outcomes by Contentiousness If an individual perceives outcomes to be much worse, and to some extent, even much better, then they are also more likely to perceive the issue as less contentious. It may be that policy actors who perceive deteriorating or improving conditions also feel that an issue may be less contentious as conditions shift from status quo and change occurs, for better or worse. The average for this que stion was 0.05 (on a scale from the majority of respondents feel that environment and public health outcomes are about the same now as they were two years ago. Conclusion The research presented here reviewed an in depth case study of the perceptions and priorities of individual policy actors involved in oil and gas development that uses hydraulic fracturing. The policy conflict surrounding oil and gas development that uses hydraulic 0 5 10 15 20 25 30 Less Contentious More Contentious Environment and Public Health Much Worse Worse About the Same Better Much Better

PAGE 58

49 fracturing is relatively cont entious and there are several individual traits and cognitive characteristics that are associated with this position. Policy actors feel that outcomes, in regard to environment and public health, have slightly worsened over the past two years. The issue is a priority for these policy actors and they associate hydraulic fracturing with perceived threats, particularly personal threat perceptions. In terms of general traits, there are not statistically significant differences in education, stance, political po sition, and experience with some caveats appear to be entrenched in their positions and threatened by the actions of those with whom they disagree. For sustainable environmental policy, the implications of these patterns in cognitive characteristics are sizeable. Policy actor perceptions matter for making effective and sustainable policy that governs the extraction of oil and gas using hydraulic fracturing, while si multaneously protecting land and resources. Basic characteristics such as education, stance, or political ideology do not appear to be influential in determining issue contentiousness, but rather, the results of this study indicate that in general indi vidual cognitive characteristics are more influential. Those who are threatened by the views of those with whom they disagree, have less collegial relationships with adversaries, and feel the issue is a high priority are also likely to be associated with a perception that hydraulic fracturing is a contentious issue. For policymakers and interested citizens, it appears to be important to moderate the levels of threat surrounding an issue and to encourage collegial relationships among opposing groups, in orde r to develop a policy area that is less contentious and consequently, more amenable to long lasting and equitable solutions.

PAGE 59

50 There are limitations to this study. As discussed earlier, any sampling procedure that attempts to identify a population that is n ot known apriori will encounter problems of sampling bias. We attempted to correct for that limitation by vetting the sampling procedure and final sample frame with issue experts, who felt that our population adequately represented the spectrum of individu als involved. A related limitation is that of selection bias. Respondents who chose to answer the survey may over represent extreme positions as they might be more motivated to share their opinions owing to their strongly held convictions on the issue. The rather moderate means on many questions would indicate that this is not an issue, but the risk remains and should be considered in drawing conclusions about a broader population of policy actors. Finally, any survey faces issues of survey instrumentality. The design and execution of a survey has been shown many times to be influential in responses. While we endeavor to present an unbiased survey instrument and to avoid all indications of preference, one respondent still felt should have been focused on shale oil more broadly. There is merit to this argument, as the policy issue of extraction of unconventional oil and gas has likely moved beyond just the techni que of hydraulic fracturing, but the importance of studying this high conflict policy issue remains high. The potential impacts of hydraulic fracturing on land, water, air, and other natural resources are significant. Existing policy regulating hydraulic f racturing is fractured at best and lacks coherent best practices. Unlike some other environmental policies, the development of oil and gas extraction is an issue area that garners attention nationwide, from high level lawmakers to everyday citizens. Those who live within proximity of industry operations inordinately bear the majority of the risks and harms associated with hydraulic fracturing, but the benefits are

PAGE 60

51 diffuse and enjoyed by the entire population of the US, which presents significant policymakin g challenges. The results of this study illuminate how individual traits and cognitive characteristics shape high conflict policy issues. The data and analysis gathered here offer a rare insight into the personal belief structures of policy actors and how those belief structures shape issue perceptions. Theoretically, this study advances efforts to explain beliefs and preferences of policy actors involved in high conflict policy issues. Practically, this study provides insight into how policy actors involve d in high conflict policy issues may differentiate themselves on perceptions of the level of contentiousness of a policy issue. Environmental policy is wrought with high conflict issues, and it is imperative that we continue to explore how these issues tak e shape.

PAGE 61

52 CHAPTER II I A CONFLICT WILD: EXPLAINING ACTIVITIES OF CONFLICT EXPANSION AND CONTAINMENT Understanding how people choose to participate in policy making processes is crucial to our understanding of policy outcomes, particularly in contentious policy debates. One of the primary political strategies in contentious policy debates is to expand or contain conflict (Schattschneider, 1960) . In high conflict policy debates, the strategy to expand conflict is important because more attention can be brought to the issue and more individuals can become in volved, both of which can overcome the friction found in any political system. At the same time, for those wanting to defend the status quo or avoid the possible ramifications from the conflict, the strategy is conflict containment (Baumgartner, 1989; Pralle, 2006; Schattschneider, 1960) . Despite the ubiquity of these two strategies, within the policy process literature, relatively little is known connecting conflict expansion or contain ment with different forms of participation also known as political behavior. Past research has emphasized two factors that can shape political behavior: sector (Brady et al. 1995; Carter 2016; Irvin and Stansbury 2004; Kam 2012) and individual level beliefs (Mutz 2002, 2009; Olofsson et al. 2018; Weible and Heikkila 2017) . This research considers how both sectoral affiliation and individual beliefs influence participation decisions regarding engaging in different activities to expand or contain the scope of conflict by asking: What drives conflict expansion or containment in contentious policy debates ? To understand how individuals make part icipation decisions in contentious policy settings, this research draws theoretical inspiration from policy process theories, specifically the

PAGE 62

53 Advocacy Coalition Framework (ACF). As a theoretical framework, the ACF is appropriate for this study owing to it s focus on the individual as the foundation of policymaking. According to the ACF, individuals form coalitions with the intent to impact the policy process, albeit in different ways (Sabatier & Jenkins Smith, 1993) . Individuals make intentional choices about how to influence policy over extended periods of time. Although the ACF recognizes the role o f context as a modifier of political behavior (Weible & Ingold, 2018) , empirical work most often focuses on the role of individual beliefs and disregards the influence of organizational level context. Working under the ACF and drawing on the theoretical logic as found in Schattschneider and others , this research argues that individuals make str ategic choices about their activities within the policy process to either expand or contain conflict. The study explores the different ways in which individuals choose to participate in a contentious policy debate, the use of hydraulic fracturing as an ext ractive technique for unconventional oil and natural gas. The setting of this theory testing case study (George & Bennett, 2005) is the use of hydraulic fracturing in Colorado. Hydraulic fracturing is a technique used to extract natural gas and oil from unconventional sources. The technique has grown markedly in its usage in the past two decades, which has been the source of cont entious debate in the United States and worldwide (Evensen et al. 2017; Heikkila et al. 2014; Olofsson et al. 2 018) . The data for the study is taken from surveys collected in Colorado in 2015 and 2017. Individuals identified as active participants in the policy process were asked how they participate in the policy process and how effective those activities were in achieving their goals related to oil and gas development that uses hydraulic fracturing. Although the survey sample was directed towards individuals ostensibly already involved in the policy process, or at least professionally attached to the issue, th ere were wide disparities

PAGE 63

54 in the extent of participation among these individuals and their perceptions of the effectiveness of those activities. Factor analysis revealed distinct patterns of participation related to conflict expansion and containment, and ordinary least squares regression was then used to uncover drivers associated with these patterns of participation. Results indicate that there is marked variation among patterns of participation, and that differences in participation patterns can be expla ined in part by differences in sectoral affiliation and individual beliefs. Theoretical Background Citizen participation is a vital part of a functioning democracy. In an open, transparent democracy, participation is crucial to claims of representativenes s and accountability, particularly in a pluralistic political system. Much of the literature on participation considers political participation, such as voting for representation in elections. The focus of this research, however, is on what happens between elections via participation in public policymaking. Public policy, as defined by Dye, encompasses both government action and inaction what governments choose to do as well as not to do (2002) . Within public policy, participation can be influential in problem definition and solution devel opment (Sobeck, 2003) . Participation can provide input for decision makers; it can bring about learning and new information perhaps even policy c hange and is fundamental to democratic norms 2010) . Bishop and Davis undertook a thorough review of the extant literature surrounding participation in the policymaking process and concluded that the typologies and frameworks disjointe d and in need of empirical testing ( 2002) . They argued that theorizing about

PAGE 64

55 through empirical research (2002: 14). Ten years prior in some of the first empirical studies on participation in th e policy process, Roberts and King used grounded theory to explore how individuals active in the policy process might decide to participate, and if so, how (1991) . We cannot understand the impact of participation without an awareness of how and why indiv iduals participate in the policy process, yet there is a lack of solid empirical evidence that explains what influences participation differentiation among stakeholders. As Cairney notes, understanding the why people make decisions, we need to know (2016: 24 25) . Knowledge of political participation in policymaking is inhibited by the under development of, one, the conceptualization of political participation and, two, theory explaining its variation. This paper contributes both. Conceptualizing Political Participation: Expanding & Containing Conflict Assuming that direct democracy or involvement in all government decisions is not possible for all citizens, Schattschneider argued that participation in policymaking is shaped by exis ting power structures (1960) . Policy change is often the result of altering the scope of conflict; therefore, it is strategically important to expand and contract the scope of c onflict within policymaking (Pralle, 2006) . Schattschneider theorized that individuals, who are otherwise locked out of the policy process or lack enough power to change it, expand the scope of conflict using a variety of strate gies to attract sympathizers and allies (1960) . For example, in an attempt recrui t individuals new to the issue to support their side, and/or shift the issue to new decision making venues.

PAGE 65

56 Many of the frameworks and theories of the policy process pull theoretical insights from Schattschneider to address conflict expansion and containm ent (see Baumgartner & Jones 1993; Kingdon, 1984; Olson, 1965) . The logic of Schattsc hneider is applied widely, but without adequate or encompassing conceptualization. The common focus has been on venue shifting as the mechanism with relatively less work on seeking new allies (see work by Baumgartner et al. 2009; Baumgartner & Jones 1993; Jones & Baumgartner 2012, for example) . There is limited work in the ACF regarding conflict expansion or containment. Empirical work drawing upon participation in ACF research underemphasizes political engagement in that research most often explores participation as a determinant in the formation and maintenance of advocacy coalitions (Ingold and Varone 2012; Nohrstedt 2011; Weible et al. 2011) or as an explanation for policy change (c.f. Albright 2011) . This research focuses on neither; rather, it focuses on the intermediary step of political engagement. Pralle undertook one of the most comprehensive studies to date on participation via conflict expansion and containment in environmental policy settings, but focused solely on advocacy groups (2006) . She rec ognized how the dueling roles of conflict expansion and containment shape the policy debate and outcomes. Her work revealed that the activities of individuals involved in policy advocacy are not random; they are the result of strategic and intentional choi ces made to influence policy. This paper adopts the basic conceptualization offered by Pralle and expands upon it by drawing inspiration from Goertz and Diehl (Goertz 2006; Goertz and Diehl 1993) . In a seminal work on the constructs of endurin g rivalries, Goertz and Diehl noted that there is a continuum of constructs, not absolute polarity (1993). Conflict expansion is similar in that we should expect shades of gradation when developing constructs of conflict expansion and containment via activ ity choice. We should see patterns of activity choice

PAGE 66

57 among policy actors associated with goals of conflict expansion, or conversely, conflict containment. There has been some empirical work that can generate limited expectations regarding the types of ac tivities that are related to conflict containment or expansion. Past empirical work on conflict containment has demonstrated that important activities are related to countering arguments you disagree with and collaborating with the opposition to limit the scope of the conflict (Pralle, 2006) . By dampening the conversation through either countering opponents or working with opponents surrounding a policy debate, the dominant coalition can maintain the status quo and contain the de bate. Similarly, strategic brokering has been shown to be effective in containing conflict because it keeps the discussion within a single venue, thus benefitting the dominant, pro status quo coalition (March & Olsen, 1984) . Government agencies can also serve as a player in the policy processes, especially in implementing (hence supporti ng existing policies) or providing an opportunity for different political voices to be heard (e.g., through established administrative practices in rulemaking). With the government as a player itself in policymaking (Denhardt & Denhardt, 2000) , sharing information and opinions with the government may also be important in maint aining the current direction of policy and, hence, containing the scope of the conflict. By commanding the conversation within the government, the dominant coalition can defend their jurisdiction (Hoberg & Phillips, 2011) Mobilization has long been considered the primary activity for conflict expansion (Downs, 1957; Schatt schneider, 1960) . Schattschneider observed that increasing the number of allies in a coordinated effort can be significantly influential in expanding the scope of conflict n in a conflict (2006: 24) . Baumgartner and Jones also found empirical

PAGE 67

58 support for the tactic of coordinating allies with the intent to expand the conflict (1993) . The media is often seen as a tool of conflict expansion, by changing the prevailing view of the issue (Baumgartner et al. 2009; Olofsson et al. 2017; Shanahan et al. 2011) . Political participation constructed via activities individuals use to engage in the policy proce ss is thus a continuum from containment focused efforts, such as brokering or countering arguments, to efforts directed at expanding the scope of conflict, like mobilization or communicating with the media. Individuals may partake in some or all of these a ctivities, at differing levels of effectiveness, in order to advance their goal of conflict expansion or containment. Explaining Political Participation: Sector and Individual Influences The field of public policy still has remnants of its behavioralist fo undations by emphasizing the individual as the driver of political phenomenon. In doing so, however, the field overlooks the influence of organizational level factors that enable and constrain political behavior. This paper incorporates both by identifying hypotheses that emphasize organizational level factors (i.e., sector) as well as individual level factors. Past research, albeit disjointed and in need of refinement, has shown that there are a multitude of ways in which individuals can choose to particip ate in either conflict expansion or containment, and those decisions are driven in part by sectoral affiliation and individual beliefs. Sector Under the banner of conflict expansion, Real Dato undertook a detailed theoretical synergy of four major policy p rocess frameworks regarding their respective interpretations of the influence of context as they relate to strategic participation in the policy process (2009) . He asserted that context is a major determinant of participation decisions and called for empirical

PAGE 68

59 applications exploring how context impacts stakeholder behavior within the policy proces s. There are contextual nuances to participation that can inform decision making within the policy process. Irvin and Stansbury argue that context needs to be accounted for in order to design participatory practices that are appropriate for the community u nder consideration (2004) . Context influences factors, both internal and external, that may impact opportunities and conditions for participation. Rydin and Pennington agree that participation in the policy process should be grounded by exploring the incentive structures of the involved individuals to better understand how and why those individuals choose to participate, and what the resulting outcomes might be (2011) . Jones and McBeth claim that individuals make strategic choices abou t their activities in the policy process, with intent to influence the conflict (2010) . Individuals are motivated in different ways to participate in the policy process, and there is a lack of empirical research that identifies how those incentive structures might vary across contexts (Nohrstedt, 2013) . Most research cites resources as the primary determinant in participation decisions (Hessels and Terjesen 2010; Pfeffer and Salancik 2003) . Many scholars have called for participation studies to move beyond consideration of resources alone in determining participation decisions and to explore other ways in which participation decisions may be influenced (Brady et al., 1995; Irvin & Stansbury, 2004b) . Empirical research within the ACF literature recognizes context as a modifier of action, but typically at the level of political system (Fischer 2014; Leifeld 2013; Nohrstedt and Olofsson 2016; Sabatier 1998) . In an effort t o develop a framework of strategic political participation by (1999: 825) . Hillman and Hitt also found that participation was in fluenced differently by organizational level characteristics than individual -

PAGE 69

60 level characteristics. There has been some work at the organizational level (see Crow & Baysha, 2013; Weible & Sabatier, 2005) , but there remains a lack of empirical studies. This research mo ves beyond political context to explore how organizational context as sectoral affiliation is related to participation decisions. The limited work that has been done connecting activity choices in a contentious policy debate to conflict expansion has onl y shown that generally, individuals will undertake strategic action in order to meet their goals of conflict expansion or containment (Baumgartner and Jones 1993; Kingdon 1984; Schattschneider 1960) . Individuals participate in the policy process according to whether they f ind themselves on the winning or losing side of the debate. In the state of Colorado, unconventional oil and gas extraction using the technique of hydraulic fracturing is currently occurring and has been for many years. The industry is comprised solely of companies in the private sector, who could be considered the winning side of the debate as their goal extract resources using hydraulic fracturing is currently allowed. Since extraction is happening, the industry and its partners dominate the question of whether hydraulic fracturing should be allowed. What remains now are largely questions related to how the extractive activity should be regulated. State and local governments provide the regulatory support for the activity to occur (Baumgartner & Jones, 1993) . Following past empirical work on conflict expansion, it stands to reason, therefore, that private and public se ctor context will be associated with the dominant coalition and consequently, conflict containment. Hypothesis 1a: Private and public sector will be associated with conflict containment. fracturing to occur, but most effort on this side of the debate is focused on regulatory questions

PAGE 70

61 surrounding how the process should be regulated. This side of the confl ict could be considered the losing side, comprised almost exclusively of organizations and individuals in the nonprofit sector in Colorado. In an effort to expand the scope of the conflict, as past work on conflict expansion has shown to be the aim of the losing side of a policy debate (Baumgartner & Jones, 1993) , individuals in the nonprofit sector, therefore, should be a ssociated with patterns of conflict expansion. Hypothesis 1b: Nonprofit sector will be associated with conflict expansion. Individual Beliefs The ACF highlights the importance of beliefs as a uniting mechanism in the formation of coalitions (Fischer 20 14; Sabatier and Weible 2007) regarding a policy issue will motivate their behavior to seek out other individuals who share the same beliefs and desire to participate in the policy process. The framework views the policy process as the interaction of competing advocacy coalitions, within which individuals share beliefs (Sabatier & Jenkins Smith, 1993) . Extensive ACF empirical work has considered the role of beliefs within the policy process, as this assumption is fundamental to the framework (Fischer 2014; Ingold and Varone 2012; Weible 2005) . Within the ACF literature, it is generally accepted that deep core beliefs are fairly resistant to change; there may be more flexibility and hence opportunity for learning or policy change within policy core beliefs and mo st certainly in secondary beliefs (Weible & Sabatier, 2006) . It has been argued that the framework needs more empirical work on understanding how beliefs are differentiated, to what end that purpose serves, and t he implications of beliefs on individual activity (Nohrstedt & Olofsson, 2016a) . If beliefs and their

PAGE 71

62 translation into policy are fundamental to coalition activity, it stands to reason that there should be a connection between beliefs and participation decisions. Foundati onal work from March and (1984: 737) . There is a multitude of conflicting evidence on the role of individual beliefs in influencing participation decisions. This leads to conflicting expectations. One of the goals of this research is to bring clarity to the empirical knowledge regarding the influence of individual beliefs on participation decisions. In prevailing theory on conflict expansion, it has been demonstrated repeatedly that the dominant coalition in a policymaking process will prefer to contain the conflict and the non dominant coalition will endeavor to expand the scope of the conflict. In Colorado, where hydraulic fracturing is currently occurring, the dominant coalition should be comprised of individuals who hold a pro s tance on hydraulic fracturing and will be associated with conflict containment. This is hypothesis 2a. Relatedly, individuals who hold an anti stance on hydraulic fracturing will be associated with conflict expansion. This is hypothesis 2b. Hypothesis 2a: Pro stance will be associated with conflict containment. Hypothesis 2b: Anti stance will be associated with conflict expansion. Past research has shown that it is imperative that policy actors feel their actions may have a viable chance of influencing the process (Gray, 1989; Lubell, 2004; Sowa, 2009) . A lack of viable options might lead to a lack of activity. Classic policy theory regarding conflict expansion focuses on the role of viable venues. Individuals who are on the winning side of a policy conflict are more likely to perceive readily available and receptive venues that are making decisions to maintain the status quo and contain the conflict. Those on the losing side may conversely feel a

PAGE 72

63 lack of viable options and may pursue activities to expand the scope of conflict through alternative, curren tly non viable venues instead (Holyoke et al. 2012; Pralle 2003) . This leads to the third hypothesis, which argues that perceptions of viable venues will be associated with conflict containment. Hypothesis 3: Positive perce ption of viable venues will be associated with conflict containment. Finally, perceptions of threat can be influential in participation decisions (Cobb & Elder, 1972; Heikkila & Weible, 2017) . Threat is often assumed to be an underc urrent of motivation for all policy actors in a given policy process. Schattschneider and Downs both recognized that threat is particularly useful for the losing side as it attempts to expand the scope of a conflict in the hunt for policy change (Downs, 1957; Schattschneider, 1960) but other work, particularly in the context of high conflict policy debates, argues that both sides use threat (Jaspal & Nerlich, 2014) . In the context of hydraulic fracturing, Jaspal and Nerlich demonstrated that threat is by pro fracking a nd anti (ibid.: 359). The final hypothesis predicts that increasing perceptions of threat will be motivating for both conflict expansion and conflict containment. Hypothesis 4: Perceptions of threat will be associated with both conflict expansion and conflict containment. Figure 1 illustrates the logic of the model of participation choice: sectoral affiliation and individual beliefs drive activity choice, which falls into patterns described by either conflict expansion or containment. This forms t he dependent variable of the model: patterns of participation. There are several ways in which an individual can participate in the policy process,

PAGE 73

64 and this research argues that 1) there are patterns to those activity choices and 2) the patterns conform to either conflict expansion or containment. Sector and individual beliefs are the independent variables of interest. Figure 1: Determinants of Participation in the Policy Process Model Specification The analysis draws upon a set of surveys first administered in Colorado in 2015 and repeated in 2017. The intent of the survey was to understand stakeholder perceptions of and priorities regarding the politics and regulations of hydraulic fracturing. Surve y questions were inductively developed after in depth interviews with key stakeholders and deductively driven by prior theory. The survey was reviewed in its entirety by an advisory board of influential and experienced individuals who are highly engaged in the topic of hydraulic fracturing. The sample

PAGE 74

65 for the survey was purposive. Individuals directly or indirectly active in the policy debate were identified through internet research and snowball methods from interview responses and advisory board recommend survey sample is not intended to be generalizable nor a reflection of public opinion; it is intended to be a survey of the perceptions and attitudes of individuals who are already engaged in the policy debate to varying extents. The survey builds on previous survey work done in New York, Texas, and Colorado. In total, 1004 people were administered the survey and 421 responded, for an overall response rate of 41.9%, which is acceptable for an online survey (Nulty, 2008) . The data is a pooled cross sectional set. 7 In 2015, 453 people were administered the survey and 235 responded for a response rate of 51.9%. In 2 017, 551 individuals were administered the survey and 186 responded for a response rate of 33.8%. The unit of analysis is the individual. Some predictors are measured at the organizational level, but these are considered to be influences upon the individua Dependent Variable The survey asked respondents to what extent they engage in and use effectively eight different activities to achieve their personal or profession goals related to oil and gas development that uses hydraulic fracturing: Brokering agreements between parties ( Brokering ), Countering arguments made by people you disagree with ( Countering Arguments ), Mobilizing the public ( Mobilizing ), Collaborating with people you disagree with ( Collaborating ), 7 The survey data is not a panel set. The sample was recreated each time to maximize the identification of individuals active in the issue of hydraulic fracturing in Colorado. A total of 64 individuals responded to both surveys.

PAGE 75

66 Coordinating political activities with allies ( Coordinating ), Providing information to government officials ( Government Information ), Providing information to the news media ( Media ), Sharing your opinion with government officials ( Sharing Opinion ). 8 These activities were vetted in interviews and with the advisory board to represent a comprehensive spectrum of participation. Responses were measured on a Likert type scale from 0 (Not engaged) to 1 (Engaged, but n ot effectively) to 2 (Engaged and moderately effective) to 3 (Engaged and very effective). In absolute terms, providing information to government officials was deemed to be the most effective activity for respondents to achieve their goals, followed by coo rdinating political activities with allies and brokering agreements between parties. Generally, if respondents were engaged in an activity, they reported moderate levels of effectiveness. Across responses and activities, the mean level of effectiveness was 1.45, between engaged but not effectively and engaged and moderately effective. Providing information to government officials had the highest average level of effectiveness (1.80), followed closely by sharing your opinion with government officials (1.72) and countering arguments made by people you disagree with (1.68). The average level of effectiveness for the remaining activities was much lower, with mobilizing the public reported to be the least average effective activity. The effectiveness scale of th e responses is key to conflict expansion and containment. As discussed in the previous section, developing constructs for conflict expansion and containment based on patterns of activity choice should be built upon a continuum, not a stark distinction (Goertz, 2006) . For example, individuals hoping to expand the scope of conflict may participate in most if not all of these activities, but there is variation in the perceived levels of effectiveness. 8 See Appendix B for the full version of all relevant survey questions.

PAGE 76

67 This is wher e factor analysis becomes most effective to discover latent constructs within the scaled responses. Latent construct analysis through exploratory factor analysis (EFA) was used to determine if there are patterns among choice between activity types. Latent construct analysis aims to uncover underlying constructs among related items. It is often used in survey research when researchers want to summate or condense responses related to an overarching topic to reveal underlying patterns within responses (c.f. Finkel, 1985; Perry, 1996) . 9 Initial factor analysis revealed two primary factors. 10 Orthogonal va rimax rotation was used to distill the identified components. Best practices for break points in social scienc es were followed (see Appendix C for Stata output of exploratory factor analysis). 11 A scree plot of the eigenvalues of the factors against the mea n was used to confirm the appropriate number of factors to keep was two (Cattell, 1966) . The total variance explained between the two factors is 59.12%. Table 1 presents the factor loadings for each item. Table 1: Rotated Factor Loadings by Item ITEM ROTATED FACTOR LOADINGS Brokering agreements between parties 0.7034 Countering arguments made by people you 0.6126 9 EFA can be used as an analytical tool in and of itself (Costello and Osborne 2005; Rummel 1970) or it can be used as an intermediary step in a larger analytical process, often to create independent or dependent variables that are more amenable to analysis (Santos, 1999) . The latter approach is used in this research. Some researchers claim that using summated scales through latent construct analysis is more valid than using one response as an indicator of a broader concept (Rummel, 1970) ; however, many also note that use of latent construct analysis is a combination of science and art, simultaneously deductive and inductive . 10 Kaiser Meyer Olkin (KMO) is a measure of sampling adequacy, estimating the commonality among the items as appropriate for factor analysis. KMO measures for all items were above 0.70, which is acceptable for this research (Kaiser, 1974) factor analysis. 11 Costello & Osborne note that due to the nature of social science data, traditional break points utilized in physical science research for retaining factors should be relaxed (2005). Therefore, two factors with eigenvalues above 1.0 were retained in this a nalysis and all components with rotated factor loadings above 0.60 were included (Kaiser, 1958) .

PAGE 77

68 disagree with Collaborating with people you disagree with 0.6775 Providing information to government officials 0.7421 Sharing your opinion with government officials 0.7286 Mobilizing the public 0.8648 Coordinating political activities with allies 0.8648 Providing information to the news media 0.6973 The first factor Containing grouped Brokering agreements between parties, Countering arguments made by people you disagree with, Collaborating with people you disagree with, Providing information to government officials, and Sharing your opinion with g overnment officials. The eigenvalue for this factor was 3.50. This factor appears to capture policy actors working to contain or mitigate the scope of conflict by maintaining the status quo and conforms to past empirical work regarding patterns of activity choice related to conflict containment. The second factor Expanding grouped Mobilizing the public, Coordinating political activities with allies, and Providing information to the news media. The eigenvalue for the second factor was 1.23. These policy actors appear to be more motivated towards expanding the scope of the conflict. These results conform to anecdotal observations of variation in type of participation and effectiveness among engaged policy actors and reveal that there is a distinct pattern to activity related to conflict containment as well. Factor loading scores for each factor were then predicted for all observations. Factor scores represent a likelihood, from negative (less likely) to positive (more likely) of the respondent conforming t o the response patterns of a given factor. The distribution centers around zero, with a standard deviation of 1. These scores comprise the dependent variable (continuous) for the OLS regression models.

PAGE 78

69 Independent Variables There are four independent vari ables: sector, stance, viability, and threat. 12 The sectoral affiliation of each respondent was recorded as either public, private, or nonprofit. The sample includes 70 respondents in the nonprofit sector, 134 respondents in the private sector and 217 respondents in the public sector. Anti is a dichoto mous variable that reports stance as the number of respondents who would prefer to completely stop or limit the use of hydraulic fracturing. About one third of respondents reported an anti stance on hydraulic fracturing. The reminder of the sample reported a pro stance to continue at current levels or expand markedly the use of hydraulic fracturing. Two independent variables capture the individual beliefs of respondents: Viability and Threat . Viability tests whether respondents feel that their actions to i mpact the policy process may have the potential in a receptive venue to actually achieve their goals related to hydraulic fracturing. For this dichotomous variable, respondents were asked whether there are any organizations or individuals who have the auth ority and trust to negotiate policy solutions to oil and gas issues in Colorado. 221 respondents, or nearly two thirds of the sample, agreed that there are viable options to negotiate policy solutions. Threat is measured in two ways. Respondents were asked to rank on a Likert type, five point scale the extent to which the views and actions of those with whom they disagree with on oil and gas development that uses hydraulic fracturing threaten them personally or professionally ( Personal Threat ) or the state of Colorado ( State Threat ). The scale ranged from not at all (1) to a great deal (5). On average, feelings of personal threat were reported markedly lower at 2.95 12 See Appendix D for basic statistical description of the dataset.

PAGE 79

70 compared with feelings of threat against the state of Colorado at 3.67. The distribution of r esponses for personal threat was much more evenly distributed than for threat against the state of Colorado, for which the responses were amassed at the far end of the scale. Three control variables were used in the analysis. Education is a discrete nomin al variable measuring the highest level of education attained. The sample is relatively highly were asked how much of a Priority it is to deal professionally or personally with political and policy issues related to oil and gas development that uses hydraulic fracturing. The five point, Likert type scale ranged from not at all (1) to the highest priority (5). The mean was 3.77, indicating that this issue is a rel atively high priority for most respondents. Finally, Year was included for year fixed effects to control for any within year variation in the sample. 13 First order pairwise correlations did not reveal any concerns for collinearity among variables. The final model is: Participation = Sector + Viability + Threat + Anti + Education + Priority + Year Results Table 2 presents descriptions for variables of interest by participation type: Containing versus Expanding. Means or modes on variables of interest were calculated to describe potential relationships among participation and sectoral affiliation and individ ual beliefs. Note that variables are measured on differing scales and are not continuous, therefore direct interpretation or comparison is not possible. However, comparisons among magnitudes are valid. Table 2: Variable Description by Participation Type 13 There is no prior for year fixed effects to affect the DV.

PAGE 80

71 Variable (scale) Factor 1: Containing Factor 2: Expanding Sector (mode) Public Nonprofit Stance (Pro/Anti) Pro Anti Viability (0/1) Yes Yes Personal Threat (1 to 5) 2.79 3.26 State Threat (1 to 5) 3.40 3.84 Education (Increasing) 5.38 4.91 Priority (1 to 5) 3.94 4.02 The mode is presented for sectoral affiliation. Among respondents who conform to the Containing type of participation, the most common sectoral affiliation is public and prevailing stance is pro. Among those respondents who can be characterized as Expanding participation type, the most common sectoral affiliation is nonprofit, but only marginally mo re so than private sector affiliation, and the prevailing stance is anti. Both types of participants believe that there are viable means for impacting the policy process. Threat, both personal and against the state of Colorado, is on average higher for Exp anders but threat against the state of Colorado is higher than personal threat, regardless of participation type. Education is slightly lower among Expanders , and below the full sample mean as well. The issue appears to be a relatively similar priority for both types of participation. In summary, Expanders are primarily nonprofit sector actors who are slightly more threatened by the issue of hydraulic fracturing than their private and public sector counterparts, who are generally working to contain the scop e of conflict.

PAGE 81

72 OLS Regression Models An OLS regression was run multiple times to determine if there are any significant relationships with sector and individual level characteristics: first with factor loading scores for Factor 1: Containing as the depende nt variable and again with factor loading scores for Factor 2: Expanding as the dependent variable, then separately with each individual activity component that feeds into the factors. Tables 3 and 4 present the results of the regressions with the factor f ollowed by each of its components. Table 3: Regression of Sector and Individual Characteristics on Factor 1: Containing and Components 14 Variables Factor 1: Containing Component: Brokering Component: Countering Arguments Component: Collaborating Component: Government Information Component: Sharing Opinion SECTOR Public 0.388*** (0.148) 0.045 (0.180) 0.234* (0.132) 0.150 (0.151) 0.188 (0.137) 0.249* (0.131) Private 0.207 (0.170) 0.042 (0.209) 0.015 (0.147) 0.187 (0.164) 0.033 (0.158) 0.034 (0.155) INDIVIDUAL Viability 0.435*** (0.112) 0.372*** (0.129) 0.235** (0.094) 0.335*** (0.107) 0.391*** (0.109) 0.375*** (0.106) Personal Threat 0.055 (0.047) 0.033 (0.055) 0.005 (0.043) 0.051 (0.045) 0.054 (0.046) 0.052 (0.043) State Threat 0.014 (0.059) 0.001 (0.067) 0.019 (0.055) 0.071 (0.056) 0.071 (0.058) 0.085 (0.058) Anti 0.366*** (0.122) 0.328** (0.148) 0.138 (0.115) 0.046 (0.119) 0.311*** (0.114) 0.163 (0.119) Education 0.106** (0.042) 0.081** (0.047) 0.086*** (0.033) 0.095** (0.042) 0.006 (0.043) 0.007 (0.040) Priority 0.425*** (0.071) 0.466*** (0.077) 0.223*** (0.061) 0.220*** (0.066) 0.345*** (0.070) 0.333*** (0.072) 14 The component models w ere run as both an OLS and an ordered logit regression given the ordered nature of the dependent variable (from 0 = not engaged to 3 = engaged and very effective). Statistical significance was the same in both models. OLS coefficients are shown here to enh ance interpretation.

PAGE 82

73 Table 3 2015 0.006 (0.102) 0.020 (0.119) 0.098 (0.088) 0.085 (0.103) 0.100 (0.099) 0.134 (0.100) Constant 2.335*** (0.439) 0.914** (0.443) 0.402 (0.393) 0.101 (0.407) 0.051 (0.425) 0.104 (0.409) Observations 309 316 315 313 316 317 R squared 0.240 0.171 0.161 0.111 0.154 0.130 Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.10 Table 4: Regression of Sector and Individual Characteristics on Factor 2: Expanding and Components Variables Factor 2: Expanding Component: Mobilizing Component: Coordinating Component: Media SECTOR Public 0.845*** (0.133) 0.662*** (0.163) 0.705*** (0.158) 0.589*** (0.143) Private 0.478*** (0.149) 0.274 (0.181) 0.148 (0.173) 0.776*** (0.161) INDIVIDUAL Viability 0.137 (0.099) 0.277** (0.117) 0.233** (0.117) 0.189* (0.108) Personal Threat 0.007 (0.044) 0.004 (0.048) 0.040 (0.055) 0.002 (0.049) State Threat 0.144** (0.059) 0.132** (0.063) 0.245*** (0.070) 0.005 (0.063) Anti 0.361*** (0.115) 0.489*** (0.137) 0.161 (0.131) 0.149 (0.130) Education 0.062 (0.039) 0.127*** (0.043) 0.004 (0.048) 0.025 (0.041) Priority 0.266*** (0.063) 0.235*** (0.082) 0.356*** (0.068) 0.361*** (0.069) 2015 0.000 (0.096) 0.038 (0.109) 0.095 (0.113) 0.044 (0.106) Constant 0.899** (0.368) 0.497 (0.434) 0.693 (0.443) 0.132 (0.429) Observations 309 316 316 316 R squared 0.336 0.263 0.292 0.205 Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.10

PAGE 83

74 Sector appears to have a significant impact on participation decisions. For Containers , being in the public sector is positive and statistically significant, indicating that public sector respondents, all else equal, are more likely to participate as Conta iners than their counterparts in the private and nonprofit sectors. Further, although the variable Public is statistically significant for Expanders , the direction of the coefficient is negative, indicating that public sector respondents are less likely to participate as Expanders , all else equal . The same holds for private sector respondents; policy actors in the private sector are statistically significantly less likely to participate as Expanders , albeit at half the magnitude of effect than Public , compa red to respondents in the nonprofit sector. In the context of hydraulic fracturing in Colorado, this is a confirmation of Hypothesis 1a conflict containment is associated with the private and public sectors and Hypothesis 1b conflict expansion is ass ociated with the nonprofit sector. In addition, holding a pro stance is statistically significant but negatively correlated with conflict expansion, and conversely an anti stance is positive and statistically significantly correlated with conflict expansio n. These results confirm hypotheses 2a and 2b. In terms of individual characteristics, Viability was positive and statistically significantly related to Containing . Respondents who feel that there are viable options for impacting the policy process are in creasingly likely to participate as Containers . The same did not find statistical significance for Expanding. This finding is intuitively valid. The activities that load on Containing brokering agreements, countering arguments, collaborating across coali tions, informing government officials and sharing opinion with government officials are activities that need viable venues or receptive decision makers in order to achieve success. The fact that viability is significant for Containers suggests that those mobilizing may feel locked out of the system and are invested in expanding the scope of the conflict to gain access to viable venues.

PAGE 84

75 This is a confirmation of the third hypothesis. Personal Threat was not statistically significant in either model. Howev er, increasing perceptions of threat against the state of Colorado were positive and statistically significant, all else equal, for Expanders. Increasing perceptions of threat against the state of Colorado appear to be motivational in prompting activities that comprise Expanding : mobilizing the public, coordinating allies, and communicating with the media. This is a partial confirmation of the fourth hypothesis. Component Analysis Factor analysis is often criticized as blunt as a tool for measuring attitudes (Beavers et al., 2013) . What do we pick up or loose when using factor analysis instead of using the components themselves? Best practices of factor analysis recommend comparing how correlations among factor scores compare to correlations among the components . When considering each type of activity choice directly with sector and individual level characteristics, some in teresting insights appear. OLS regression for Factor 1 indicated that being in the public sector was positive and statistically significant for individuals participating as Containers. However, component analysis revealed that public sector affiliation is only statistically significant for two of the five factors, and with a negative relationship for one of those factors ( Countering Arguments ). This indicates that policy actors who deem countering arguments to be a highly effective activity are not likely to be in the public sector. This runs counter to some prior work on conflict expansion in environmental policy settings (see Pralle, 2006) . Another departure from the results of the Facto r 1: Containing is found within Anti

PAGE 85

76 stance. Anti was found to be strongly significant but indirectly related to Containing , indicating that all else equal, individuals holding an anti stance on hydraulic fracturing were less likely to participate as Conta iners and more likely to participate as Expanders . Again, though, only two of the five components display this same pattern: Brokering and Government Information . Anti was not statistically significantly associated with Countering Arguments , Collaborating , or Sharing Opinion . In the state of Colorado, it could be argued that those who hold a pro stance on hydraulic fracturing are the current dominant group. Dominant groups generally prefer containment (Baumgartner and Jones 1993; Pralle 2009; Schattschneider 1960) , and therefore the negative sign on the overall factor coefficient makes sense. However, there are some distinctions in terms of the components themselves. The overall regression results from Factor 2: Expanding indicate that individuals in the nonprofit sector are most likely to participate as Expanders . As noted earlier, individuals in the private sector almost all hold a pro stance on hydraulic fracturing and could be perceived as the dominant coalition in Colorado, where hydraulic fracturing is currently occurring. However, the strong indirect relationship between Private sector individuals and Expanding is driven entirely by the factor Media . Here the relationship is again negative, which makes intuitive sense that individuals in the private sector and on the dominant side are unlikely to seek communication with the media, and the same holds for individuals in the public sector. The role of threat is an interesting finding across all components and factors. For those seeking to expand the scope of the conflict, State Threat is motivating to act, but in terms of individual factors, the relationship holds only to Mobilize or Coordinate . If part icipation in the the media, which is common in numerous media studies surrounding hydraulic fracturing, State

PAGE 86

77 Threat would not have appeared as a statistically sig nificant relationship. This finding provides clarity to previous work on the role of threat in high conflict policy debates. Much of that work asserts that powerful opposition by those with whom you disagree should be threatening and therefore motivation f or action (Pralle, 2006) . The results here demonstrate that threat may not be the motivator it is assumed to be, particularly when considering the goals of conflict expansion versus containment. Conclusion This research set ou t to understand variation in levels of participation in contentious policy debates by exploring the drivers of those patterns. There is a lack of research within the policy process literature that identifies how participation within the policy process may reflect the scope of conflict. The field lacks a solid conceptualization of conflict expansion as political engagement and relies too heavily on individual explanations, discounting organizational level influences such as sector. This research fills that g ap by first, conceptualizing conflict expansion via activity choice and second, explaining the drivers of activity choice as influenced by both organizational and individual level variables. Findings indicate that there are specific activities related to either conflict expansion or containment, and individual choice among those activities is associated with sectoral affiliation and individual beliefs. This research is an important step in understanding which activities may be more effective or accessible to certain groups or policy goals. This is important in understanding structural barriers to participation and has implications for democratic participation and representativeness.

PAGE 87

78 For organizational level influences, the first hypothesis predicted that the private and public sectors would be associated with conflict containment and the nonprofit sector would be associated with conflict expansion. This was confirmed in the results. For individual level influences, in Colorado, the dominant side currently holds a pro stance on hydraulic fracturing and the opposition holds an anti stance. The second hypothesis predicted that a pro stance would be associated with conflict containment and an anti stance would be associated with conflict expansion. This was als o demonstrated in the results. The third hypothesis predicted that positive perceptions of viable venues through which to control the policy process would be associated with conflict containment, and this was confirmed. Finally, the fourth hypothesis was p artially confirmed in that it expected to find beliefs of threat associated with both conflict expansion and containment, but findings revealed a statistically significant relationship between threat and conflict expansion only. Too much of the public poli cy literature emphasizes individual factors over organizational. This paper shows both are important and shape the type of political activities people engage in (Weible & Carter, 2017) . There are theoretical implications for the ACF in these findings. While much current work simply assumes that coalitions form and are maintained through s hared beliefs, this research reveals that there is more going on. Individuals may be incentivized towards or constrained against participating, despite their beliefs, based on their context. This research corroborates much of what has long been suggested a bout the motivations for participation in the policy process: there is intentionality behind the activity choices that individuals make to influence policy, and this has implications for theory about coalitional politics in contentious policy debates. In C olorado, the dominant side holds a pro stance on hydraulic fracturing and is primarily

PAGE 88

79 comprised of private and public sector individuals. The anti stance side is currently largely comprised of individuals in the nonprofit sector. There is high collinearit y in this particular study between stance and sector, and that is encouraging for the validity of the findings here. However, we can imagine circumstances in which the pro stance would not be the dominant coalition, for example, settings with moratoriums a gainst hydraulic fracturing like New York. In those places, the pro stance coalition would work to expand the conflict and the results would be flipped: pro stance should be associated with conflict expansion. We could also imagine a scenario in which ther e is not yet a dominant coalition in new policy debates in which case there might not be a significant relationship at all between stance and pattern of participation. A reasonable question raised by this research is whether conflict expansion is poss ible in the public sector. The results here would indicate it is unlikely; however, remember that the survey question was not dichotomous participate or not but rather gauged the effectiveness of a given activity. Public sector policy actors may in fac t engage in conflict expansion activities but feel that those activities are less effective in terms of conflict expansion. The institutional conditions of the public sector may prevent effective conflict expansion, and this remains a point for further res earch. Relatedly, this research emphasizes the importance of developing a construct for understanding the relationship between activity choice and conflict expansion or containment. Relying on the past empirical work exploring specific activities in confl ict expansion would approach. For conflict containment, past findings would suggest that countering arguments made by persons with whom you disagree would be a suffici ent measure of conflict containment. However, the results here demonstrate that this approach is not only insufficient, it actually

PAGE 89

80 masks some of the more interesting nuances to individual level determinants of participation. There are limitations to this research that point to next steps in advancing research on conflict expansion and containment. First, using factor analysis to create the dependent variable dictates the nature and the distribution of the outcome measure. There are a limited number of appl ications that take this approach to first determine patterns within an outcome variable then use those patterns as a dependent variable (see Oser, Hooghe, & Marien, 2012) . The non ordered nature of nominal categories may call into question the choice of OLS regression as an estimation technique; ordering the types of activities by frequency, for example, might result in estimates that lend themselves to more intuitive interpretation. However, this approach masks the tendency of individuals to participate in regularized patterns and with differing values to different ac tivities. An individual may only mobilize supporters once a year, registering as infrequent activity on many traditional measures of policy activity, but may feel that one effort is highly effective. The differentiation in activity weighting is captured th rough this research, which is a strength of the study. Another limitation of this study is the selection bias of the sample, which constrains generalizability. While the study purposively sampled engaged stakeholders, it is surprising that there are many respondents who participate infrequently in the policy p rocess. This could be an artefact of the survey design and the questions asked. Relatedly, another limitation of the study may be self selection bias. There could be an overrepresentation of active participants who were more likely to answer the participat ion question, creating a systemic difference on the missing data observations. This requires further analysis to determine if the differences are indeed systematic and generating bias in the results or rather the result of the particular setting.

PAGE 90

81 Finally, there may be barriers to participation in some activities. Some policy actors may perceive a lack of access in order to share their opinion or information with the government. Others may feel that they have no means of collaborating with those with whom th ey disagree or countering arguments made by those with whom they disagree. This, however, is one of the primary motivations for and contributions of this work: to uncover systemic inequalities in the system perceived or actualized and consider how thos e inequalities can be addressed. The identification of distinct patterns of participation as conflict expanding or containing in nature is a solid contribution to the policy process literature. In addition, uncovering the influence of sectoral affi liation and individual beliefs on those patterns of participation is a first step towards better understanding how multiple contextual factors interact to influence participation decisions. More work remains to distill further influencers and continue to d evelop theories of participation in the policy process, so that we might better understand how contentious policy debates develop. The quality of political debate suffers when participation is marred by systemic obstacles. In the immortal words of Charlott e (1846 ) .

PAGE 91

82 CHAPT ER IV STRATEGIC MOTIVATIONS IN POLICY GOVERNANCE Research in environmental governance has been generally characterized by collaboration studies. Most studies are done at the organizational level and focus on the structural characteristics of the network. Far fewer studies consider the individual network actor and how individual choices influence the resulting network. Most networks involve both choice and chance in formation. This research endeavors to explain some of the choice. When making net work choices, individuals face a selection problem. Owing to costs of networking and limited capacity for response from advocacy targets, individuals are forced to select partners strategically in a network space characterized by competition (Lee et al., 2012) . There have been several studies of individual influence on strategic policy network choices (Fischer & Sciarini, 201 3; Hadden & Jasny, 2017; Johari et al., 2006; Watts, 2001) , but few studies consider how the context of policy actors or their individual perceptions influences their policy network choices. A theory of individual choice must account for two mechanisms: an external mechanism to account for the alternatives on offer and internal mechanism to account for choice between alternatives. Sectoral context as the external mechanism sets the menu of options available, and individual characteristics as the internal mechanism determine choice among those options. This research aims to uncover how individuals choose to build relationships by asking: How do sectoral context and individual perceptions influence the structure of policy networks ? The analysis will model t he incentives behind network choices and clarify network behavior by not just asking with whom an individual collaborates but also by relationships is a departure f rom many network studies and is a novel contribution.

PAGE 92

83 This study focuses on policy actors, who are individuals identified as active in the policy issue and are distinct from the general public. The research assumes that policy actors make intentional choi ces regarding their network connections, and these choices are constrained by sectoral context and individual characteristics. The objective of this research is to identify variation in policy networks and how this variation can be explained by sectoral co ntext and individual characteristics, through the lens of the Advocacy Coalition Framework (ACF). The ACF purports that policy issues are characterized by competing advocacy coalitions, which can be understood as networks of individuals (Matti & Sandström, 2013) . This research will employ a relational approach (Ingold, 2011) to explore the policy networks of individuals involved in unconventional oil and gas development in Colorado. By adding an element of perception via importance, this research uncovers how individuals value relations hips in networks. While the ACF recognizes the fundamental role of individual choice in network behavior, it lacks theory regarding how those network choices might be influenced by sectoral context and individual characteristics. The paper contributes to e nvironmental governance and ACF literatures by interacting the sectoral context with individual characteristics to develop individual level choice models explaining network preferences. The research will apply theories of organizational management, behavio ral economics, and network science to assess the self interested and benefit maximizing choices of individuals on network structure. Literature Review Individual political behavior is constrained by time, financial resources, capacity, and most importantl y, institutional rules and norms that vary by sector. Scharpf summarized the a shorthand term for organizational capabilities and the rules governing their employment w ill constrain,

PAGE 93

84 (1989: 149) . He goes on to distinguish opt rules. The institutional context captured in this research as sectoral context creates a setting within which some engagement options are impossible or at l east severely disadvantaged. Sectors exhibit fundamentally different dynamics. The private sector can be characterized by (Lynn, Heinrich, & Hill, 2001: 33) . Governance arrangements are embedded in larger contexts and those contexts have implications for the execution of governance arrangements. For exa mple, there are regulations and norms constraining the actions of nonprofit and public sector employees in regards to their efforts to impact the policy process. Recent work has also demonstrated that the influences on individual choices in relationship bu ilding are contingent upon context (Nohrstedt & Bodin, 2019) . Owing to a combination of sectoral context, individual cognitive limitations, and basic capacity constraints, policy actors are forced to make strategic decisions and target specific individuals or organizations to achieve their policy goals (Fischer & Sciarini, 2013) . The policy networks that arise may be driven by these constrained decisions. Policy networks are the interactions of individuals and organizations with an aim to achieve personal or professional goals and can be viewed as the aggregation of many individual level strategic ch oices (Gulati, 1999) . Policy networks are constantly changing owing to contextual influences and policy (Fischer & Sciarini, 2013) . Policy actors may maximize their limited capacities by targeting those organizations and individuals whom they feel would have the highest potential impact on the policy process.

PAGE 94

85 Several scholars recognize the importance of policy networks for power, legitimacy, resources, information, and advocacy (Agranoff & McGuire, 2001; Hill & Lynn, 2005; Lubell, Milward, 2001) . Much of the empirical policy network research focuses on collaborative networks under the guise of governance or collective action studies. Governance networks arise as a means of alternate service delivery (Ansell & Gash, 2007; Jing & Besharov, 2014; Provan & Milward, 2001; Wood, Don na & Gray, 1991) , more inclusive government (Scott & Thomas, 2017) , dealing with complexity Donna & Gray, 1991) , and generating positive outcomes (Provan & Kenis, 2008; Provan & Milward, 2001) . These stu dies have focused on the structural characteristics of the networks and how networks are determinants of outcomes. Governance studies often recognize the role of constraints and competition within an advocacy space (see Lee, Lee , & Feiock, 2012) , but fail to advance a full network theory of the policy process as it fails to explain the characteristics of the components of a network the individual (Dowding, 1995) . There appears to be some level of strategic awareness among individuals and/or organizations of the importance of intentional decision making regarding networking building. Lee et al. explored the structural features of an interorg anizational network to answer questions surrounding the emergence of policy networks, the forces that shape network formation, and how networks evolve over time (2012). The authors found that organizations seek to build stable and sustainable relationships , which require more than simple exchange transactions. Past work in policy networks, particularly in regards to policy change, have shown that network partners are at least somewhat aware of the advantages to be gained through strategic tie building (Chalmers, 2013; Granovetter, 1985) . Leifeld and Schneider recognized the role of political opportunity

PAGE 95

86 structures and transaction cost analysis in individual decision making regarding whom polic y actors choose as contacts (Leifeld & Schneider, 2012) . Using the case of information exchange, the authors determined that policy actors knowingly attempt to minimize transaction costs and maximize returns within their constrained capacity. This paper builds upon this finding by capturing the strategic choices of policy actors in network connections, as influenced by sectoral context and individual characteristics. This research also moves a step beyond previous network studies by exploring not only with whom a policy actor chooses to connect but also the level of importance that policy actor assigns the interaction. By giving a measure of importance, the policy actors reveal how they value that relationship by re porting the utility of that interaction, which can yield deeper insight about strategic network choices. We learn more about relationship building in governance settings. With the individual as the unit of analysis, as opposed to the majority of studies do ne at the organizational level, this research can explore the institutional opportunities, constraints, and incentives in combination with individual characteristics that influence strategic network udy of public administration and policy, extant research on (interorganizational) networks has been focused on how networks work and how networks help individuals or organizations perform better based on macrolevel een directed to microlevel analysis of the emergence (2012: 567 568) . The authors note that hat follow a different analytical strategy and try to come up with analytical strategies that match with

PAGE 96

87 (Snijders, Matzat, & Reips, 2012: 2) . This current project end eavors to lend insight this empirical puzzle. Conceptual Framework A core assumption of the ACF is that individuals come together in coalitions to influence policy (Weible, Sabat ier, & McQueen, 2009) . The ACF argues that individuals hold beliefs regarding policy; those beliefs have been shown to be influential in motivating coalition development (see Henry, Ingold, Nohrste dt, & Weible, 2014a; Matti & Sandström, 2013; Weible, 2005) unclear, howe ver, whether those beliefs alone are sufficient to bond advocacy coalitions (Henry et al., 2014b; Schlager, 1995) , and significant empirical work has identified networks as a potential tool to better understand coal itions and coalition development (Fischer et al., 2012; Ingold, 2011; Lienert et al., 2013) . This research does not dispute that shared beliefs can encourage the formation of policy coalitions as networks but rather aims to bring more theoretical clarity to that assumption in the ACF by testing whether tie formation might be motivated by context as well. For example, there is substantial empirical evidence that belief homophily is a driver of network formation (Jenkins Smith et al., 2014; Metz, Leifeld, & Ingold, 2018) ; this research seeks to explore how other contextual v ariables, specifically sectoral affiliation, may also influence relationship building. The goal of this research is to identify and explore political behavior as networks that reflect whom an individual policy actor targets to achieve his or her policy go als. These are related to but may also be distinct from policy coalitions. This study recognizes that coalitions exist as part of a larger policy network but is not explicitly concerned with coalition

PAGE 97

88 identification. The policy networks in this study are n ot intended to represent or define coalitions 15 , but rather will be differentiated by sectoral context and individual characteristics. 16 Network science has shown that there is much to be gained theoretically by considering network structure, moving beyond position or belief driven coalition structure (Bodin, 2017; Jackson, 2005) . In this study, policy actors holding opposing stance on a policy issue may find themselves in the same policy network when conditioned on institutional context such as sector. The ACF is a framework that enables multiple theoretical argu ments (Jenki ns Smith et al., 2014) , and while conforms to ACF assumptions. There is overlap in this question compared to what is traditionally asked in ACF coalition stu dies. This research expands knowledge regarding network relationships, supplementing theoretical and empirical ACF findings with knowledge gained in behavioral economics, organizational management, and network science. This study models the incentives to foster valuable relationships and brings clarity to network behavior. By focusing on the individual instead of the organization, the research reveals a different perspective on governance and how individuals value relationships in policy networks. Early em pirical work within the ACF by Zafonte and Sabatier found that although beliefs were important in understanding coalitions, beliefs alone were not enough to explain coordination (Zafonte & Sabatier, 1998) . Scholars have continued to build on this foundation recognizing that there are in terdependencies between network structure and belief systems (see Henry, Lubell, & McCoy, 2011; Schneider, Scholz, Lubell, Mindruta, & Edwardsen, 2003; 15 There are several reasons why this particular network analysis does not intend to define n or represent coalitions. First, the network is unidirectional and not symmetrical, in that the ties are not necessarily reciprocal. Second, the network is also bipartite, or two mode, composed of two different types of vertices, in this case individual org anization. Finally and relatedly, the network is incomplete, as is common in social network analysis. 16 It should be acknowledged that s ome network studies have defined and i dentified coalitions using the type of network described proposed here . There is a sizeable literature within the ACF that debates how coalitions should be defined: by beliefs, or by network connections, or by shared venues, among others.

PAGE 98

89 Weible, 2005) . While there has been significant work exploring the connection between belief systems and coalition structure, there remains a gap in ACF literature regarding network valuation, spe cifically tie valuation. First, the focus has been on structural features of the network, such as transitivity or reciprocity (cf. Scott & Thomas, 2017) , and second, extant lite rature has failed to take into account how context might interact with belief systems to influence network structure. Leifeld and Schneider found that the context of the target organization can impact the network choices of policy actors (2012) . In 2011, Ingold bui lt upon previous studies recognizing the role of individual beliefs in networks by studying if cooperative beliefs result in collaboration (Ingold, 2011) . The author argued that although beliefs were shown to be instrument al in predicting network cohesion, the settings may constrain or enable network ties as well. Homophily has been found to be a strong determinant of ties (Calanni et al., 2015; Lee et al., 2012; Leifeld & Schneider, 2012; Mcpherson, Smith Lovin, & Cook, 2001) . However, this stream of research f ocuses on beliefs as the driving force of homophily and often misattributes the social mechanisms underlying network structure to social influence when strategic partner selection may be more impactful (Lewis, Gonzalez, & Kaufman, 2011; Steglich, Snijders, & Pearson, 2010) . The approach in this study recognizing the strategic tie behavior of policy a ctors extends the sphere of homophily to the sectoral context, which lends insight into how opposing actors may engage in similar tie building strategies. When tie behavior on networks is intentional, as it arguably is in policy advocacy networks, then we can turn to organizational management and network science for empirical direction on expectations, complemented by behavioral economics. Theories of behavioral economics adopt tools of economic analysis to explore, for example, political issues. Behaviora l

PAGE 99

90 economics assumes that individuals act strategically and in their self interest to maximize their utility (Downs, 1957; B. D. Jones, 2001; Shepsle, 2008) . These assumptions are not limited to economic choices; this applies political behavior in choice of network targets as well (Currarini, Jackson, Pin, & Papadimitriou, 2010) . Revealed preference theory in economics suggests that when a consumer chooses one product over another, we learn decision making structures (Sen, 1971) . Similarly in a network context, we learn something making structures when he or she reveals their preferences through choice among target organizations in an advocacy setting. Theoretical economic models of individual strategic choice in network structure have been developed (see Bloch & Jackson, 2006; Galeotti, Goyal, & Kamphorst, 2006; Goyal & Vega Redondo, 2005; Jackson & Watts, 2002; Jackson & Wolinsky, 1996) but empirical applications are lacking. Individuals have limited relationship budgets, as Burt argues in his network theory of social capital (2000) . Forming and maintaining network ties carries considerable uncertainty and transaction costs (Lee et al., 2012) and this uncertainty is linked to context (Nohrstedt & Bodin, 2019) . In empirical applications, the cost and associated risk of forming links have been shown to be a major de terminant in strategic network formation (Berardo, 2014; Goyal & Vega Redondo, 2005) . As the cost of forming links rises, fewer links will be formed. In addition, under conditions of constraint, individuals who perceive a policy issue to be contentious appear to be motivated form what they hope will be high quality links (Ansell & Gash, 2007; Gray, 1989) . Issue contentiousness seems to play a role in individual decision making regarding relationship building, but that role is not identical across sectors. If we assume that the cost of building relationships is relatively similar across sectors, then it stands to reason that issue contentiousness will play a larger role in relationship building in sectors with fewer resources,

PAGE 100

91 such as the nonprofit or public sectors (Hessels & Terjesen, 2010; Miller Millesen, 2003) . Guo and Acar identified the conditions unde r which nonprofit organizations are more likely to collaborate and found that resources are a major determinant in network structure (2005) . The authors emphasize, however, that resource dependency and transaction cost theories alone are not sufficient explanations for strategic network choice context matters as well. H1: Perceptions of issue contentiousness will be relatively more influential in relationship building among nonprofit policy actors than among public and private sector policy actors. network ef forts is influential in their determination of whether a relationship will be formed and maintained subsequently referred to as viability (Ansell & Gash, 2007; Bradford, 1998; Geoghegan & Ren ard, 2002; Rogers et al., 1993; M. Schneider et al., 2003) . Extensive work in network science has considered how to build a viable network and the traits that encourage network efficiency and stability, yet there remains a gap in knowledge at the indivi dual level that under different sectoral contexts. Collaborative governance scholars recognize that inter sectoral networks are crucial to the delivery of public services (Hill & Lynn, 2005; Milward & Provan, 2000; Provan & Milward, 2001) . Foundational work in collaborative governance established that perceptions of viability are integral to public sector organizations embedded in governance arrangements (Provan & Milward, 2001) . This research inherently recognizes that there are differences in sectoral affiliation among network participants but has not yet directly addressed how those sectoral differences may enable or constra in network choices. Past work in nonprofit research has shown that before even engaging in network activity, nonprofit organizations first consider potential for benefits (N icholson Crotty, 2009) . Sowa demonstrated that perceptions of

PAGE 101

92 rewards from collaboration are strongly influential in promoting collaborations for nonprofit organizations (2009) . Turning to the private sector, substantial work in organizational theory has demonstrated that individuals aim to build a competitive advantage by building key relationships to leverage their efforts (Burt, 2009; Hallen & Eisenhardt, 2012; Wu, Shih, & Chan, 2009) . Seminal work in collaboration from Ansell and Gash noted that all individuals are more likely to build relationships when they believe that those relationships will yield results (2007) . Bringing together these disparate streams of work, the second hypothesis asserts that all individuals regardless of sectoral affiliation will be motivated by perceptions of viability to build relationships. However, there should be variation in th e strength of the effect. H2: Perceptions of viability to impact policy will be influential in relationship building for all respondents. Acquiring and maintaining resources is crucial to the survival of the private sector, and this can bias behavior decisions (Pfeffer & Salancik, 1978) . Research in organizational management has found that under conditions of uncertainty, organizations in the private sector prefer to strengthen already existing ties (Beckman, Haunschild, & Phillips, 2004; Staw, Sandelands, & Dutton, 1984) . Related to sensemaking (Weick, 1995) response indicates that when perceiv ing threat or risk, private sector individuals seek stability in their policy networks. Staw et al. demonstrated a tendency in private sector organizations towards valuing existing structures when under conditions of threat (1984). The research considered learned or have more to

PAGE 102

93 lose under conditions of threat (Beckman et al., 2004) and may be more likely to value existing relationships, prompting the third and final hypothesis. H3: Perceptions of threat will be more influential in relationship building for private sector actors than among nonprofit or public sector policy actors. Methodology This study is set in the policy issue of unconventional oil and gas exploration using the technique of (Arnold, Long, & Gottlieb, 2017; Rabe & Borick, 2013) . Although the use of hydraulic fracturing is not new to Colorado, its usage has grown markedly in recent decades and ha s brought the issue to the forefront of energy policy in the state (Heikkila & Weible, 2017) . Colorado serves as a typical or representative case in the issue of hydraulic fracturing. Typical case studies are often used to confirm causal expectations (Seawright & Gerring, 2008) , which is the aim of this study. The explanatory nature of this research is well suited for caus al confirmation. Colorado meets a basic standard of representativeness, as a comparison for other states engaged in hydraulic fracturing, owing to the public visibility of the issue combined with active engagement by both sides of the issue yet relatively less economic dependence on the extracted resource. Data is drawn from a survey first administered in 2015 and repeated in 2017. The survey sample used a combination of online research, snowball sampling, and expert knowledge to identify individuals active (see Heikkila, Weible, & Pierce, 2014; Nohrstedt & Olofsson, 2016; Weible, Olofsson, Costie, Katz, & Heikkila, 2016) . Sampling for the survey was purposive. This study aims to explain the network decisions of individuals involved in the po licy process, not the general public. As such, the survey was intentionally directed to individuals

PAGE 103

94 identified as policy actors. The total dataset is comprised of 421 responses. In 2015, the survey was sent to 453 policy actors with 236 responses, for a re sponse rate of 52.1%. In 2017, the survey was administered to 551 policy actors with 185 responses, for a response rate of 33.6%. Both response rates surpass a reasonable target for online research (Nulty, 2008) . Social Network Analysis The objective of this research is to capture the interactive effects of sectoral context and individual characteristics on network choices. Social network analysis is therefore an appropriat e tool for analysis not only in its descriptive capacity, but also in its extension into inferential explanations (Ingold & Pflieger, 2016) . Specifically, this study will rely on Exponential Ran dom Graph Models (ERGM). ERGMs are well suited for this study, which explains the variation in network structure between sectors. In an ERGM, the dependent variable is the network itself; the reference distribution reflects the observed network. The model captures the regularities in the network and analyzes whether the structural features of the observed network are more or less than we might expect by chance (Anderson, Wasserman, & Crouch, 1999; Kriv itsky, 2012) . An ERGM explains network structure; the basic assumption of the model is that the chance of a tie between two actors is dependent on the properties of those two actors as well as the other relationships that the actors may have. The observ ed network is the result of a process, however noisy it may be, that is quantifiable and testable. ERGM is preferable to other inferential network specific network (Desmarais & Cranmer, 2012: 431) . The stochastic approach of ERGM is appropriate for confirmatory studies such as this one. Operationalization

PAGE 104

95 The dependent variable in this study is a policy network. Policy networks are the networks of organizations and individuals that are accessed by policy actors to achieve their personal or professional goals related to oil and gas development that uses hydraulic fracturing. Respondents were given a battery of fifteen organizations Fede ral Government , State Government , County Government , City Government , Oil and Gas Industry , Oil and Gas Professional Associations , Environmental or Conservation Groups , Real Estate Developers or Home Builders , Agricultural Organizations or Farmers , Organized Citizen Groups , Churches or Other Religious Organizations , Universities or Colleges , Consulting Firms or Think Tanks , Informal Personal Networks , and News Media and asked to rank importance of interaction on a Likert type scale from not at all important (1) to very important (4) with an opti on for N/A (refer to Appendix B : Full Survey ). The list method of the survey yields 15 ties formed for each respondent. The tie weight is given as the reported importance of the interaction, resulting in an a ctor organization (two mode) bipartite network with directed ties. Institutional context will be captured through the sectoral affiliation of the policy actor. Several individual characteristics will be assessed through independent variables and controls. For the first hypothesis, the explanatory variable is perceptions of issue contentiousness. Respondents were asked to rank the level of political contention about shale oil and gas development in comparison to other contentious political issues in Colorado on a Likert type five point scale ( variable of the second hypothesis is perceptions of viability to impact hydraulic fracturing policy. Respondents were asked whether there we re any individuals or organizations who have the authority and trust to help negotiate policy solutions to shale oil and gas in Colorado (yes/no). Finally, the independent variable for hypothesis three deals with perceptions of threat.

PAGE 105

96 Respondents were ask ed to rank on a Likert type, five point scale (1 to 5) whether the views and actions of those you disagree with on shale oil and gas development threaten 1) them personally or professionally and 2) the state of Colorado. Both measures will be tested. Contr ols for this study include issue stance and level of priority personally or professionally to deal with policy issues related to shale oil and gas development. Results The entire policy network regarding hydraulic fracturing in the state of Colorado is com prised of 365 respondent nodes (missing removed) with directed ties towards 15 target organizations. The network is a two mode bipartite actor organization network in which ties are neither cooperative nor reciprocal. 17 To disentangle drivers of network beh avior and to capture how sectoral context interacts with individual characteristics to manifest as network choices at the individual level, it was necessary to parse the whole network into three sectoral policy networks, driven by the sectoral affiliation of the respondent. As with other work exploring strategic relationship building (Berardo & Scholz, 2010; Henry et al., 2011; Nohrstedt & Bodin, 2019) , the premise is that by comparing network structures as differentiated by sectoral context, we learn more about the underlying individual choices. Following Gould and Fernandez (1989), the respondents were parsed by into three non overlapping subgroups corresponding to their individual sectoral affiliation: public, private, and nonprofit . Figure 1 illustrates the three sectoral networks. The entire policy network is not of central analysis to this research, but rather three separate networks differentiated by the sectoral affiliation of the respondent. In this manner, the research capture s sectoral influence by holding it constant in each network. For example, in the 17 For a similar approach, see Metz, Leifeld, and Ingold, 2018.

PAGE 106

97 network and Figure 1. Sectoral Policy Networks Each network is comprised of the same 15 target organizations towards which ties of differing weight assigned by each respondent are directed. A tie is formed when a respondent ass igns a value to the interaction. For example, a respondent in the nonprofit sector may assign a tie weight of 1 towards Federal Government , a tie weight for 4 for Environmental or Conservation Groups , N/A for Real Estate Developers , and so on for all fifte en organizations. This network mapping procedure results in directed, non reciprocated ties, or edges, from respondents towards the previously identified fifteen target organizations; explaining the variation in tie weighting the assignment of importance of interaction is the aim of this research. 18 The sectoral networks that comprise the dependent variables are formed by ties that are both within network and outside network. This is where this research departs from most 18 Malang, Brandenberger, and Leifel d employed similar logic in developing two mode networks representing the veto events of parliamentary decisions in the European Union (2017) .

PAGE 107

98 governance studies, which focus o n cross sectoral relationships. While important, this research focuses instead on the unidirectional tie weighting from the alter (respondent) to the ego (target organization). Some of these ties are intrasectoral and some are cross sectoral. The key is th at the alters in each network are the same but each ego weights those interactions differently, giving a utility or value to the interaction. This is, by definition, the tie weight but observed through reported data instead of generated through network mea sures, as is more common. Table 1 details some of the characteristics of the sectoral networks. The public network is comprised of 182 respondents. The mean importance score assigned to target organizations is Ranking the mean importance score by target organization, public sector respondents felt that interactions with State Government , followed by Industry , and County Government were most important to achieving their personal or professional goals related to the policy issue of hydraulic fracturing. The least important interactions were with Consultancies , Informal Networks and lastly Churches , which received the lowest mean importance score of all organizations in all networks at 0.81. Table 1. Sectoral Policy Networks: Description Network Nodes Mean Importance (0 4) Most Important Targets (Mean) Least Important Targets (Mean) Public 182 2.12 1. State Government (3.35) 2. Industry (3.00) 3. County Government (2.88) 13. Consultancies (1.43) 14. Informal Networks (1.39) 15. Churches (0.81) Private 123 2.18 1. State Government (3.37) 2. Industry (3.23) 3. County Government (2.71) 3. Professional Associations (2.71) 13. Consultancies (1.52) 14. Media (1.42) 15. Churches (1.06) Nonprofit 59 2.41 1. State Government (3.29) 2. Organized Citizen Groups (3.22) 3. Environmental Groups (3.14) 13. Real Estate Developers (1.67) 14. Professional Associations (1.45) 15. Consultancies (1.37)

PAGE 108

99 Descriptive characteristics for the network of private sector respondents is fairly similar to the public sector network. There are 123 respondents in the private sector network who assigned a slightly higher mean importance score at 2.18, in comparison with the public sector network. The rankings of most important organizations are the same, with marginally higher scores for State Government and Industry and a somewhat lower score for County Government . Professional Associations are tied with County Government as the third most important interaction, which is not surprising considering these are private sector respondents. Consultancies and Churches are again among the least important target organizations, but with Media replacing Informal Networks . ran king among the least important interactions for private sector respondents is unexpected considering the abundance of privately funded media campaigns regarding hydraulic fracturing. The nonprofit network diverges from the public and private networks. It i s the smallest network by far, at 59 respondents, but with the highest mean importance score at 2.41. As for the private and public networks, State Government is the most important target organization but with the lowest mean of the three networks (3.29). For respondents within the nonprofit sector, the second most important target organization is Organized Citizen Groups , an organization that was ranked ninth in both the private and public sectors. Environmental Groups , which was ranked at seventh in the p rivate sector and fifth in the public sector, is the third most important target organization for nonprofit sector respondents. This may be evidence of some institutional homophily in that respondents in the nonprofit sector are inclined to work with targe t organizations with the same nonprofit sector affiliation, perhaps to ease the transaction costs of tie formation or because these organizations may be more likely to recognize the different

PAGE 109

100 institutional circumstances under which nonprofit organizations must operate. Additional support for institutional homophily can be found among the least important target organizations in the nonprofit network Real Estate Developers , Professional Associations , and lastly Consultancies all three of which are private sector organizations. Evidence of institutional homophily may indicate structural obstacles to cross sector engagement, which is troubling in terms of equitable democratic representation, or it may instead be the result of the particular advocacy targetin g preferences of policy actors within the nonprofit sector. Modal responses within each sector and by target organization are fairly uniform across sectors and target organizations and follow the mean distribution closely, with a few exceptions in the no nprofit network. The modal response for County Government and City Government each come in at 4, but the mean for these two categories is sufficiently low enough (at 2.96 and 2.83, respectively) that these organizations are not found in the top three targe t organizations for the nonprofit sector. This indicates that there is a wider range of perceptions within the nonprofit sector regarding the importance of interactions with county and city organizations; some respondents feel those interactions are of the utmost importance, while other feel that the focus should lie elsewhere. Turning now to the independent and control variables, Table 2 presents descriptive characteristics. Means and modes are presented for ordinal variables; counts for dichotomous varia bles. On most measures, the public sector appears to approach a normal distribution with wider variation in the private and public sectors. For the first hypothesis, the issue of hydraulic fracturing is most contentious for respondents in the nonprofit sec tor, by a marked amount. The issue is least contentious for respondents in the public sector. The independent variable for the second hypothesis asks whether there are viable organizations that are able to impact policy

PAGE 110

101 regarding hydraulic fracturing. For all sectors, the majority of respondents confirmed that there are organizations who can impact policy. The public sector holds the highest percentage of respondents, at 42%, who do not feel that there are organizations capable of impacting policy compared with 29% in the private sector and 32% in the nonprofit sector. The third hypothesis tests the independent variable related to threat. By far, public sector respondents feel the least threated themselves (2.53); private sector respondents feel the most thr eatened personally (3.34). Table 2. Independent and Control Variables: Description Variable Mean Mode Issue Contentiousness Public: 3.77 Public: 4 Private: 3.99 Private: 4 Nonprofit: 4.14 Nonprofit: 4 Threaten You Public: 2.53 Public: 2 Private: 3.38 Private: 3 Nonprofit: 3.34 Nonprofit: 3 Priority Public: 3.55 Public: 4 Private: 3.93 Private: 4 Nonprofit: 4.05 Nonprofit: 4 Viability Yes No Public: 101 Public: 73 Private: 80 Private: 33 Nonprofit: 40 Nonprofit: 19 Stance Anti Pro Public: 63 Public: 105 Private: 17 Private: 99 Nonprofit: 53 Nonprofit: 6 In terms of controls, prioritization of the issue of hydraulic fracturing is lowest in the public sector and highest in the nonprofit sector. Stance is most evenly distributed in the public orrelated with sectoral affiliation for the private and nonprofit sectors. The clear majority of nonprofit sector

PAGE 111

102 stance. Exponential Random Graph Model With the structure of the sectoral policy networks established, we can now turn to the inferential analysis to explore the determinants of individual network choices. As revealed in Figure 1, the sectoral networks themselves are dense and complex, and further a nalysis is needed to understand the differences among the networks. This research uses valued ERGMs, which are a recent advancement in the network space but have grown in popularity in recent years (see Ingold & Leifeld, 2016; Scott, 2016) . Valued ERGMs take into account relationships of varying strength, rather than the standard binary on/off, and offer a more nuanced approach to modeling a network (Krivitsky, 2012) . The reference distribution for an ERGM is a probability distribution of possible networks of a set number of iterations. The observed network is modeled as if it has been drawn from this distribution, preserving the structural components of the observed network in the generation of the reference distribution. The structure is endogenous, creating a dyadic dependent model. The approximation algorithm for simulating ERGMs is Markov Chain Monte Carlo maximum likelihood procedure (MCMC), using a binomial (bounded) reference distribution owing to the multinomial, discrete nature of the dependent variable. This valued ERGM model is similar to a multinomial logistic regression; the coefficients on the covariates can be interpreted as additive log odds, generally in terms of the given edge or tie existing. ERGMs ultimately produce an estimate for various para meters, indicating how a potential tie would change the network statistic

PAGE 112

103 by one unit, which changes the log odds for the tie to exist at a given value. The conceptual model for the ERGM is: Policy Network = (Contentiousness) + (Viability) + (Threat) + Con trols The analysis was run using ergm package in R (Handcock, Butts, Goodreau, Krivitsky, & Morris, 2017; Krivitsky & Butts, 2018) . Separate valued ERG Ms were run for each network: public, private, and nonprofit. Table 3 presents the results. Table 3: Maximum Likelihood Estimation, Valued ERGMs by Sector (Standard Error 19 ) Variable Public Network Private Network Nonprofit Network Sum 2.026 ( 0.3278)*** 0.488 (0.111)*** 1.168 (0.160)*** Contentiousness 1.257 (0.080)*** 0.239 (0.076)** 0.192 (0.054)*** Viability 0.169 (0.026)*** 0.557 (0.041)*** 0.459 (0.038)*** Threat 0.826 (0.066)*** 0.137 (0.045)** 0.424 (0.064)*** Priority 0.296 (0.178) 1.594 (0.065)*** 0.278 (0.056)*** Stance (Anti) 0.178 (0.045)*** 0.070 (0.062) 0.332 (0.065)*** AIC 55783 37794 15693 BIC 55655 37678 15600 *** p<0.001 ** p<0.01 * p<0.05 ERGM, like the intercept in ordinary regression, and the term is functionally irrelevant for this study. The coefficient estimates presented in Table 3 are the change in the l og odds likelihood of a tie for a unit change in the predictor. The first hypothesis posited that perceptions of contentiousness would be relatively more influential in the nonprofit and public sectors than in the private sector. The estimates on the varia ble Contentiousness are statistically significant across all three sectoral policy networks. The sign is positive for the nonprofit sector network and negative for the private and public sector networks. All else equal, a one unit increase in 19 MCMC and goodness of fit diagnostics reported a robust model without degeneracy in all three cases.

PAGE 113

104 Contentiousness, for example fr odds of higher importance of a relationship 1.21 times in the nonprofit sector and decreases the odds of importance of a relationship to 0.78 in the private sector and 0.29 in the public sector. This indicates that perceptions of an increasingly contentious policy issue are motivational for increasing importance of relationships in the nonprofit sector but significantly lacking in the public and private sectors. This is a confirmation of Hypothes is 1. The second hypothesis tests how perceptions of viability of success resulting from relationships influence relationship building. The hypothesis suggested that perceptions of viability will be influential for all respondents. This variable is dichoto mous with a 1 indicating the respondent agreed that there are viable means to impact policy regarding hydraulic fracturing in Colorado. Viability was found to be positive and statistically significant for the public and private networks. All else equal, re spondents in the public sector who perceive viable means for influencing the policy process are 1.18 times more likely to assign higher importance to their relationships. The effect is even stronger in the private sector, at 1.75 times more likely. The non profit sector demonstrates an opposite effect, however. While still statistically significant, all else equal, perceptions of viability in the nonprofit sector are associated with lower odds (0.63) of higher relationship valuation. The second hypothesis is only partially confirmed, and there remains more work to do in this area to understand why. The role of perception of threat is the focus of the third hypothesis. The hypothesis posits that higher relationship valuation in the private sector will be asso ciated with increased perception of threat. This relationship was found to be true as the effect was positive and statistically significant, confirming the third hypothesis. While threat was positively influential among private sector respondents, the infl uence of increased threat was significantly negative for

PAGE 114

105 respondents in both the nonprofit and public sectors. In one of the strongest effects in the models, for a one unit increase in perceptions of personal threat, odds are only 0.43 in the public sector and 0.76 in the nonprofit sector that respondents will assign higher importance to relationships. The impact of the control variables is mixed. Increasing priority of issue is statistically significant but conversely related to higher value of relationshi ps. There is no relationship between priority of the issue and relationship building for the public sector, which is interesting. Even though these policy actors operate within the public sphere, priority of issue is not influential in motivating network b ehavior. As public service motivation scholars might argue, this could indicate disillusionment or desensitization (Crewson, 1997; P erry, 1996) . Holding an anti stance regarding hydraulic fracturing was positive and statistically significantly likely to increase the likelihood of a network tie among public and nonprofit sector respondents, but not statistically significant for priva te sector respondents. This may be a function of the dataset, in that nearly all respondents in the private sector hold a pro stance on hydraulic fracturing. However, that implies that there should be a converse relationship evidenced in the model, which d id not occur. Apart from Priority , the effects of the covariates in the models are diametrically opposed for the nonprofit and private sectors. When an effect is positive for the nonprofit sector, it is negative for the private sector. When the effect is negative for the nonprofit sector, it is positive for the private sector. This may be confirmation of the existing state of hydraulic fracturing policy in Colorado, in which most would consider the private sector or industry to be the winning side. The non profit sector is almost entirely composed of environmental and other groups working to limit or stop hydraulic fracturing. Theories of conflict expansion may also lend insight into why Contentiousness is motivational only in the nonprofit sector. As the

PAGE 115

106 cu compelled to expand the scope of conflict through relationship building in what they perceive to be a relatively more contentious issue. Future work should consider c ontexts in which the stance and vice versa. The overall effects of individual characteristics are more muddled in the public sector. In potential additional evidence of disillusionment or perhaps as a result of their bureaucratic nature, the public sector retreats from relationship building under conditions of contentiousness and threat, valuing relationships higher only when viable solutions are present or as dictated by professional demands. Borrowing from psychology, Contentiousness and Threat act as protective factors, while Viability acts as a risk factor. Conclusion The objective of this research was to understand and explain what motivates policy actors to value relationships in an environmental policy network. Substantial work in environmental governance has demonstrated that collaboration is often necessary, yet there is a lack of empirical evidence that moves the field forward in understanding how context affects those arrangements and ultimately outc omes (Bodin, 2017) . We must develop a better understanding of the structural conditions that influence network behavior, which is where this research contributes. The general assumptions of the s tudy posit that sectoral context is filtered through an first hypothesis tested whether perceptions of issue contentiousness would be more impactful in the no nprofit sector. Findings indicate that issue contentiousness is indeed influential upon relationship building for respondents in the nonprofit sectors, which is a confirmation of this

PAGE 116

107 hypothesis. The second hypothesis tested the influence of individual per ceptions of viability of efforts to impact policy across all sectors. Perception of viability was influential for all respondents, but negatively so in the nonprofit sector. This is a partial confirmation of the second hypothesis. The final hypothesis test ed how the views of those with whom you disagree threaten you and whether that perception is influential in motivating network behavior in the private sector. The hypothesis was confirmed; higher perceptions of threat were found to be positive and statisti cally significantly associated with increased likelihood to value network ties among private sector respondents. There are several lessons from these findings. The research brings theoretical clarity regarding the relationship between individual percepti ons and network structure, but there is more work to be done to tease out the ways in which individual perceptions influence network behavior in different sectoral contexts. The research also demonstrates that the res ponse to belief cues is muddled and tha t some individual characteristics may be more influential than others. While there has been extensive work in the field of public service motivation that explores the factors affecting motivation among public sector employees, less is known about the motiv ational structures of individuals engaged in the nonprofit sector. Given the already sizeable and growing influence of the nonprofit sector, additional research that builds upon the findings of this current project would be advised. Finally, the ranking o f State Government as the most important target organization in each network could be an indication of potential instability in the policy advocacy arena. In environmental governance settings, the capacities of both respondents and target organizations are limited. If everyone targets the same organization, in this case the state government, then the payoff of interacting with that organization goes down and consequently, the network is not

PAGE 117

108 efficient. It is similar to a collective action dilemma: the target advocacy could be considered a common poo l good in that it is rivalrous and, at least ostensibly in democratic theory, non excludable. Respondents are maximizing their own payoffs without regard for the actions of others within the network, which may be destabilizing to the system as a whole. In addition, there are likely to be inefficiencies and inequities in access and network outcomes. Nodes can only hold so many connections before they start to lose the collaborative advantage of networking (Provan & Milward, 2001) . This system is overloaded on a crucial node State Government and the potential for destabilization and i nefficiency is high. There are some limitations to this research. First, it can be questioned how much a list type network generator reflects real life connections. The list used in the survey was vetted in previous surveys and with experts, but nonetheless it is possible that important connections were omitted. In addition, as is common with survey research, it is impossible to know which specific generally. Specific individuals or organizations would be preferred to general references; however, this would render survey participation lengthy and unwieldy for the respondent. Second, the data cannot capture the extent to which there is mimicry or diffusion on th e sectoral policy networks because it is the most important advocacy target, or because targeting behavior mimics and snowballs? Finally, this research may over emphasize choice versus chance in network structure; perhaps these connections are simply not as intentional as the rational economic assumptions of this paper would imply. Despite its limitations, there are numerous contributions to this research. The fin dings yield some insights about systemic ways in which network structure may be influenced by

PAGE 118

109 sectoral context as filtered through individual characteristics. This ties the nature of individual choice to network structure, which is where this paper departs from others. Most research looks at the opposite: how structure can influence variables of interest, such as information flow. This research turns that on its head and instead looks at the determinants of network structure and illuminates some of the fund amental as sumptions about human behavior that are prevalent in environmental advocacy research. Future research on individual choice models in network behavior could lend insight into network efficiency. In a space in which so many policy actors are vying for attention from the same partners, what does this mean for network efficiency or stability? Can we compare networks across space and time to make some conclusions about network structures that lead to desirable outcomes? We can identify tradeoffs, when individual choices might outweigh (network) efficiency. Network science has theorized extensively about the formations that maximize benefits for all involved actors (Bala & Goyal, 2000; Bloch & Jackso n, 2006) , but empirical applications are lacking. Some theoretical work in network science has shown that too much pressure on one organization or individual creates an unstable and inefficient network (Golub & Jackson, 2010) , perhaps the very situation uncovered in this research. Structural Balance Theory recognizes that in g roup relationships among nodes, there exists varying degrees of connectedness that create tension within a system (Cartwright & Harary, 1956; Easley & Kleinberg, 2010) . A system is only structural ly balanced within rigid constraints. As we stray from those constraints in our observed networks, what happens to the stability of the system? This is particularly pertinent in high conflict advocacy settings in which differentiated access hinders democra (Provan & Milward, 2001: 422) ,

PAGE 119

110 and that we should not shy away from this considerable task because of the normative implications.

PAGE 120

111 C HAPTER V C ONCLUSION Dissecting political behavior is a complicated task at best. Many previous attempts to structure decision making in high conflict policy settings have fallen short of offering clarity into the inner workings of the policy actor and have often instead creat ed more confusion. This dissertation brings clarity to the structure of decision making of the individual policy actor and how context may be influential in those decisions by asking: How do context and individual characteristics influence political behavi or in contentious policy debates? This is critical research as policy debates grow ever more complex and conflictual. Democracy is fulfilled through discourse, even contentious discourse, and understanding how context influences the actions of individuals within those debates is vital to the preservation of our political system. policy conflicts generally. The research considered the beliefs and policy preferences of individuals involved at the national level in the debate surrounding the extraction of unconventional oil and gas in the United States. The focus on national level po licy actors, as opposed to the Colorado specific policy actors in the following two papers, was purposive in that national level actors should be less susceptible to local level variation and therefore the broad effects of individual characteristics should be more prevalent. Using a survey conducted in 2016 (n = 133), the research explored individual belief structures to understand their role in contentious policy debates. The research first considered whether individual perceptions of contentiousness of hy draulic fracturing were correlated with stance on the issue (pro/anti),

PAGE 121

112 political position, education, or professional experience with the issue. 20 There were no statistically significant associations between these individual characteristics and perceptions of contentiousness, which is encouraging because statistically significant correlations on these particular characteristics might indicate sample bias and/or an issue that is driven purely on political party lines. The more interesting findings came from associating other individual cognitive characteristics with issue contentiousness. Respondents who perceive hydraulic fracturing to be relatively more contentious are associated with 1) feeling more threatened by the issue, both personally and against the nation, 2) assessing their relationships with people with whom they disagree to be less collegial, and 3) finding the issue to be a high priority. Conversely, respondents who perceive the issue to be relatively less contentious are associated with 1) feel ing very little threat from the issue, 2) assessing their relationships with people with whom they disagree to be more collegial, and 3) finding the issue to be less of a priority. Finally, the research investigated the association between two outcomes e nvironment and public health with perceptions of contentiousness. In strongly statistically significant associations, it appears that individuals who perceive environment and public health outcomes to be either much worse or much better than two years pr ior also perceive the issue to be less contentious. It may be that policy actors who perceive deteriorating or improving conditions also feel that an issue loses its contentiousness as conditions shift markedly , for better or worse. 20 The majority of respondents perceive the issue of hydraulic fracturing to be relatively contentious. Respondents were asked to rate the level of politica l contention regarding oil and gas development using hydraulic fracturing, in comparison with other political issues in the United States. Response options were on a five point Likert type scale from Far less contentious (2.80% of respondents), Less conten tious (14.95%), Just as contentious (39.25%), More contentious (33.64%), and Far more contentious (9.35%).

PAGE 122

113 The second paper of the Recognizing that one of the primary political strategies in contentious policy debates is t o expand or contain conflict, the research strengthens this ubiquitous assumption with empirical evidence demonstrating that there are patterns to individual political behavior that conform to goals of conflict expansion or containment. By capturing one fo rm of political behavior as the activities that individuals choose to participate in the policy process, the research moves into the case study setting of Colorado to capture the influence of sectoral context as well individual level beliefs on individual participation decisions. The research generally argued that sectoral activities used to participate in the policy process, and those activity choices will fall into patterns that can be described as either conflict expansion or conflict containment. Using factor analysis, the research first found that there are two distinct patterns to activity choice. The first factor, conflict expansion, grouped mobilizing ind ividuals, coordinating allies, and providing information to the media as activities. The second factor grouped brokering disagreements, countering arguments, collaborating with your opposition, and sharing your opinion and information with the government a s activities related to conflict containment. These patterns to activity choice then form the dependent variable of an OLS regression to compare how sectoral context and individual characteristics influence individual activity choice. Sectoral context was captured as sectoral affiliation of the individual respondent. Individual characteristics were captured as perceptions of threat owing to hydraulic fracturing and perceptions of whether viable alternatives exist that have the potential to impact the polic y process. Findings indicate that being in the public or private sector is positive and statistically

PAGE 123

114 significantly related to conflict containment, as were perceptions of viability and holding a pro stance on hydraulic fracturing. This makes intuitive sen se in the setting of Colorado, as the status quo and regulations established by the government currently allow for hydraulic fracturing; thus those in the public sector and those who favor and participate in the extraction of natural resources using hydrau lic fracturing would be associated with conflict containment. Conversely, being in either the public or the private sector were statistically significant but inversely related to conflict expansion, indicating that the nonprofit sector is most likely to pu rsue conflict expansion. Threat against the state of Colorado and holding an anti stance on hydraulic fracturing was positive and statistically significantly related to conflict expansion, which is again intuitive. Opposition to hydraulic fracturing in the state is largely compromised of nonprofit entities who view the extractive process as highly threatening. considers yet another form of political behavior: the networking choices of individuals in this contested governance system. The research explored the influences on how the individual policy actor chooses to form relationships among different potential target organizations. Most networks involve choice and chance in for mation. This research explains some of the choice in formation. Policy actors were asked to report the importance of their relationships with 15 different organizations, with responses ranging on a Likert Public, Private, or Nonprofit based on their individual sectoral affiliation. This technique captures, and controls for, the influence of sector on the variation in reported importance of each of the 15 target organization for each individual policy actor.

PAGE 124

115 Using an Exponential Random Graph Model (ERGM), the networks then become the dependent variable and regression was run to test the relationships between individual characteristics tested in t he previous two papers: perception of issue contentiousness, perception of viability, and perception of threat. Findings indicate that issue contentiousness is influential upon relationship building for respondents in the nonprofit sectors. The perception of viability was influential for all respondents, but negatively so in the nonprofit sector. Finally, higher perceptions of threat were found to be positive and statistically significantly associated with increased likelihood to value network ties among pr ivate sector respondents but not motivational for tie valuation in the public or nonprofit sectors. What We Learned Together, these papers develop a more nuanced illustration of political behavior, conceptualized as individual participation decisions in contentious policy debates and what influences those decisions. Three themes emerged. First, increasing perceptions of issue contentiousness are associated with deteriorating relationships , heightened feelings of threat , and pressing issue priority. While this finding is not revolutionary, it is additional empirical confirmation that implies that when policy debates grow increasingly contested, individual perceptions and actions may move towards extremes. This research highlights the importance of conflict management to retain productive discourse in high conflict policy debates. Second, relationships between individuals and organizations in contentious policy exchanges are not formed randomly nor in isolation. This research has demonstrated that individual characteristics and sectoral context are influential in driving networking choices, which appear to also be made with an element of strategy in a contested policy space. Much of the extant literature on advocacy in policy making purports that relationship building is largely

PAGE 125

116 random and by chance. Additionally, much literature indicates that organizations unde r constraint are most likely to pursue as many relationships as possible to increase the likelihood of success. This dissertation demonstrates that there is rather more intentionality behind relationship formation, particularly with the intent to impact po licy and even when under conditions of constraint. The research presented here also contributes to our understanding of network stability and efficiency. The state government was decidedly perceived to be the most important target, which makes sense in a s etting where the regulatory authority is the state. However, this is concerning for the stability of the system and fair representation of all views. The capacity of the state to form relationships with all interested parties is limited, which can lead to structural obstacles for those not used to accessing the system or who do not have the power to reach decision makers. Anecdotally, this appears to be the case in that some policy actors feel shut out of the policy process. Relatedly and perhaps most impor tantly, the final theme that emerges from this dissertation is that there are structural barriers to participation. There appears to be differentiated access to activities; some activities or networks may be more effective or accessible to certain groups o r policy goals. The findings here indicate that there are a few ways in which some individuals may experience barriers to participation in the policymaking process. For some individuals, their perceptions such as threat or issue contentiousness may imp ede their activities in the policy process. The impacts of different perceptions are amplified under certain structural conditions, such as the constraints of the public sector. The institutional conditions of the different sectors either create opportunit ies or impede political behavior. Others simply feel that there are not viable options to impact the policy process. Combined, these insights indicate

PAGE 126

117 that our current system of both formal and informal participation in the policymaking process may in fact prevent fully democratic representation. Limitations Methodologically, this research faces the persistent challenges of survey research, namely the potential for sample bias and response bias. Owing to the way in which the population of policy actors was identified, there may be sample bias. However, the same sampling procedure has been used in several other research efforts with success. The population was reviewed before each iteration of the survey to affirm the ongoing activity of individuals in the po licy issue. Relatedly, there is a potential for hidden populations that were not identified through the sampling procedure. These could be individuals who are active in the policy process but not identifiable through the sampling procedure. Finally, respon se bias could be an issue in that the tend towards over representing activity. However, there is wide variation in activity levels in the responses, which in dicates that the survey did capture some individuals who have been only marginally active, as well as individuals who participate daily in the policy issue of hydraulic fracturing. Ongoing efforts to replicate the survey again in 2019 should provide additi onal insight into the role of bias in this data. In terms of case selection, while some of the research was done at the national level, the research primarily focuses on one context: Colorado. There are solid arguments as to why Colorado is a representati ve case for the policy debate surrounding hydraulic fracturing. The state has a long history with the oil and gas production and has even been using the technique of hydraulic fracturing for several decades. Although oil and gas revenue is important for th e state, which makes the issue salient, the state is not wholly reliant on these revenues, which might

PAGE 127

118 skew perceptions and risk/benefit analysis. Finally, the political ideology of the state is moderate; the debate regarding hydraulic fracturing is very m uch ongoing and does not favor anti or pro stance. In comparison to other states with more extreme debates surrounding hydraulic fracturing, such as New York or Texas, Colorado would appear to be a more moderate and representative case. However, consideri ng the role of context in this research, it is important to acknowledge that the limitations to generalizability in the findings. Is Colorado truly representative? What does this mean for the findings? The political ideology of the state may be shifting mo re liberal. What would this mean for future policymaking regarding oil and gas extraction? Some local governments have already passed moratoriums on new drilling and continue to strengthen regulations on existing production. Future research will capitalize further on the longitudinal nature of this dataset as the survey is replicated in Colorado. The same research will also be applied in a new setting Oklahoma where the history with resource extraction is also long but the political ideology is much dif ferent than Colorado. Finally, one of the strengths of this research is that it has helped develop a survey tool that can be applied in different policy areas. The findings of this dissertation are likely generalizable to other policy debates natural resou rce extraction, but not necessarily to other highly contentious policy areas. Finally, the research is both motivated by and limited by the inherent lack of confirmation regarding mental decision making structures. As with any behavioral study, it is impos sible to confirm that the findings are indeed the absolute triggers of decision making. However, the patterns discovered in this dissertation point to a routine influence of context and choice. Further research is needed to determine if those patterns hold up in similar contexts and perhaps vary in different contexts. Comparative studies will be the key to validating the findings here.

PAGE 128

119 Why It Matters By focusing on the influences of political behavior decisions, the dissertation generates knowledge about th e diversity of political behavior that might be used in different contexts and contributes to our understanding of the policy process and political outcomes. The theoretical definition of political behavior is refined by understanding the use of activities and policy networks at an individual level. Exploring distinct institutional settings with varying options for activities and political networks lends insight into why certain political behaviors emerge. Individuals filter institutional context through th eir individual characteristics, and that influences their political behavior. We can detect institutional settings and individual cognitive characteristics that shape political behavior. In addition, this research has implications for understanding content ious political exchanges. If we better understand the different toolbox that each person accesses to interact in the policy process, then we can tackle difficult issues such as disproportionate representation and unequal and inequitable outcomes. Democracy is built upon a fundamental tenet that each person if they so desire should have a voice in the process. This dissertation clarifies how people choose to exercise that voice and what determines those individual decisions.

PAGE 129

120 R EFERENCES Agency, E. P. (2016). EPA Releases Final Report on Impacts from Hydraulic Fracturing on Drinking Water . Agranoff, R., & McGuire, M. (2001). Big Questions in Public Network Management Research. Journal of Public Administration Research and Theory , 11 (3), 295 326. http://doi.org/10.1093/oxfordjournals.jpart.a003504 Albright, E. A. (2011). Policy Change and Learning in Response to Extreme Flood Events in Hungary: An Advocacy Coalition Approach. Policy Stu dies Journal , 39 (3), 485 511. Networks. Social Networks , 21 , 37 66. Ansell, C., & Gash, A. (2007). Collaborative Governance in Theory and Practice. Journal of Public Administration Research and Theory , 18 (4), 543 571. http://doi.org/10.1093/jopart/mum032 Arnold, G., Long, L. A. N., & Gottlieb, M. (2017). Social Networks and Policy Entrepreneurship: How Relationships Shape Municipal Decision Making about High Volume Hy draulic Fracturing. Policy Studies Journal , 45 (3), 414 441. http://doi.org/10.1111/psj.12175 Bala, V., & Goyal, S. (2000). A Noncooperative Model of Network Formation. Econometrica , 68 (5), 1181 1229. Baumgartner, F. R. (1989). Conflict and Rhetoric in Fren ch Policymaking . Pittsburgh, PA: University of Pittsburgh Press. Baumgartner, F. R., Berry, J. M., Hojnacki, M., Kimball, D. C., & Leech, B. L. (2009). Lobbying and Policy Change: Who Wins, Who Looses, and Why . Chicago: University of Chicago Press. Baumgar tner, F. R., & Jones, B. D. (1993). Agendas and Instability in American Politics . Chicago: University of Chicago Press. Beavers, A. S., Lounsbury, J. W., Richards, J. K., Huck, S. W., Skolits, G. J., & Esquivel, S. L. (2013). Practical Considerations for U sing Exploratory Factor Analysis in Educational Research. Practical Assessment, Research & Evaluation , 18 (6), 1 13. Beckman, C. M., Haunschild, P. R., & Phillips, D. J. (2004). Friends or Strangers? Firm Specific Uncertainty, Market Uncertainty , and Network Partner Selection. Organization Science , 15 (3), 259 275. Berardo, R. (2014). Bridging and Bonding Capital in Two Mode Collaboration Networks. Policy Studies Journal , 42 (2), 197 225. http://doi.org/10.1111/psj.12056 Berardo, R., & Scholz, J. T. (2010). Self Organizing Policy Networks: Risk, Partner Selection, and Cooperation in Estuaries. American Journal of Political Science , 54 (3), 632 649.

PAGE 130

121 Bishop, P., & Davis, G. (2002). Mapping Public Participation in Policy Choices. Australian Journal of Public Administration , 61 (1), 14 29. http://doi.org/10.1111/1467 8500.00255 Blatter, J., & Blume, T. (2008). In Search of Co va riance, Causal Mechanisms or Congruence? Towards a Plural Understanding of Case Studies. Swiss Political Science Review , 14 (2), 315 356. Bloch, F., & Jackson, M. O. (2006). Definitions of Equilibrium in Network Formation Games. International Journal of Gam e Theory , 34 , 305 318. http://doi.org/10.1007/s00182 006 0022 9 Blomquist, W. (2007). The Policy Process and Large N Comparative Studies. In P. A. Sabatier (Ed.), Theories of the Policy Process (2nd ed., pp. 261 289). Boulder: Westview Press. Bodin, O. (20 17). Collaborative Environmental Governance: Achieving Collective Action in Social Ecological Systems. Science , 357 (659), 1 8. http://doi.org/10.1126/science.aan1114 Boehmke, F. J., Gailmard, S., & Patty, J. W. (2005). Whose Ear To Bend? Information Source s and Venue Choice in Policy Making. Quarterly Journal of Political Science , 1 , 139 169. Bradford, N. (1998). Prospects for Associative Governance: Lessons from Ontario, Canada. Politics & Society , 26 , 539 573. Brady, H. E., Verba, S., & Lehman Schlozman, K. (1995). Beyond Ses: A Resource Model of Political Participation. The American Political Science Review , 89 (2), 271 294. Bronte, C. (1846). The Missionary. Paternoster Row. Bump, P. (2015, February 28). The Hi The Washington Post . Washington, D.C. Burt, R. S. (2000). The Network Structure of Social Capital. Research in Organizational Behavior , 22 , 345 423. http://doi.org/10.1016/S0191 3085(00)22009 1 Burt, R. S. (2009). Structural Holes: The Social Structure of Competition . Cambridge, MA: Harvard University Press. Cairney, P. (2016). The Politics of Evidence Based Policy Making . http://doi.org/10.1057/978 1 137 51781 4 Calanni, J. C., Siddiki, S. N., Weible, C. M., & Leach, W. D. (2015). Explaining Coordination in Collaborative Partnerships and Clarifying the Scope of the Belief Homophily Hypothesis. 2Journal of Public Administration Research and Theory , 25 (3 ), 901 927. Carter, D. P. (2016). Public, Nonprofit, and For Profit Sector Regulatory Approaches in Third Party Regulatory Administration. Journal of Public Administration Research and Theory , 26 (4), 726 744. http://doi.org/10.1093/jopart/muw022 Cartwright The Psychological REview , 63 (5), 277 293. Cattell, R. B. (1966). The Scree Test for the Number of Factors. Multivariate Behavioral Research , 1 (2), 245 276.

PAGE 131

122 Chalmers, A. W. ( 2013). With A Lot of Help From Their Friends: Explaining the Social Logic of Informational Lobbying in the European Union. European Union Politics , 14 (4), 475 496. http://doi.org/10.1177/1465116513482528 Cobb, R. W., & Elder, C. D. (1972). The Dynamics of Agenda Building. In J. Shafritz, K. Layne, & C. Borick (Eds.), Classics of Public Policy (pp. 128 134). Boston, MA: Cengage Learning. Costello, A. B., & Osborne, J. W. (2005). Best Practices in Exploratory Factor Analysis: Four Recommendations for Getting the Most From Your Analysis. Practical Assessment, Research & Evaluation , 10 (7), 1 9. http://doi.org/10.1.1.110.9154 Crewson, P. E. (1997). Public Service Motivation: Building Empirical Evidence of Incidence and Effect. Journal of Public Administration Res earch and Theory , 7 (4), 499 518. Cross, F. B. (1999). The Judiciary and Public Choice. Hastings Law Journal , 1 27. Stakeholder Understanding in Policy Making. Review of Policy Research , 30 (3), 302 320. http://doi.org/10.1111/ropr.12020 Currarini, S., Jackson, M. O., Pin, P., & Papa dimitriou, C. (2010). Identifying the Roles of Race based Choice and Chance in High School Friendship Network Formation. Proceedings of the National Academy of Sciences of the United States of America , 107 (11), 4857 4861. http://doi.org/10.1073/pnas.091179 3107 Denhardt, R. B., & Denhardt, J. V. (2000). The New Public Service: Serving Rather than Steering. Public Administration Review , 60 (6), 549 559. Desmarais, B. A., & Cranmer, S. J. (2012). Micro Level Interpretation of Exponential Random Graph Models wit h Application to Estuary Networks. Policy Studies Journal , 40 (3), 402 434. DiChristopher, T. (2016). Major EPA Fracking Study Downplayed Risks to US Water Supply, Investigation Finds. CNBC. sing Factor Scores: Considerations for the Applied Researcher. Practical Assessment, Research & Evaluation , 14 (20), 1 11. http://doi.org/10.1.1.460.8553 Political Stu dies , 43 , 136 158. Downs, A. (1957). An Economic Theory of Political Action in a Democracy. Journal of Political Economy , 65 (2), 135 150. Dye, T. (2002). Understanding Public Policy . Upper Saddle River: Prentice Hall. Easley, D., & Kleinberg, J. (2010). Ne tworks, Crowds, and Markets: Reasoning about a Highly Connected World . New York: Cambridge University Press. Choice in Large Chapter 11 Reorganizations. Cornell La w Faculty Publications , 967 1003.

PAGE 132

123 Elgin, D. J. (2014). Examining the Role of Resources, Beliefs, and Behavior in the Policy Process: A Study of Colorado Climate and Energy Politics and Policy . University of Colorado Denver. Entman, R. M. (1993). Framing: T oward Clarification of a Fractured Paradigm. Journal of Communication , 43 (4), 51 58. Evensen, D., Stedman, R., & Brown Steiner, B. (2017). Resilient But Not Sustainable? Public Perceptions of Shale Gas Development via Hydraulic Fracturing. Ecology and Society , 22 (1), 1 8. Finkel, S. E. (1985). Reciprocal Effects of Participation and Political Efficacy: A Panel Analysis. American Journal of Political Science , 29 (4), 891 913. Fischer, M. (2014). Coalition Structures and Policy Change in a Consensus Democracy. Policy Studies Journal , 42 (3), 344 366. http://doi.org/10.1111/psj.12064 Fischer, M., Ingold, K., Sciarini, P., & Varone, F. (2012). Impacts of Market Liberalization on Re gulatory Network: A Longitudinal Analysis of the Swiss Telecommunications Sector. Policy Studies Journal , 40 (3), 435 457. http://doi.org/10.1111/j.1541 0072.2012.00460.x Fischer, M., & Sciarini, P. (2013). Europeanization and the Inclusive Strategies of Ex ecutive Actors. Journal of European Public Policy , 20 (10), 1482 1498. Galeotti, A., Goyal, S., & Kamphorst, J. (2006). Network Formation with Heterogeneous Players. Games and Economic Behavior , 54 , 353 372. http://doi.org/10.1016/j.geb.2005.02.003 Geoghega n, T., & Renard, Y. (2002). Beyond Community Involvement: Lessons from Insular Caribbean. Parks , 12 (2), 16 26. George, A. L., & Bennett, A. (2005). Case Studies and Theory Development in the Social Sciences . Cambridge, MA: MIT Press. Gerring, J. (2007a). C ase Study Research: Principles and Practices . Cambridge, MA: Cambridge University Press. Gerring, J. (2007b). What Is A Case Study Good For? Case Study versus Large N Cross Case Analysis. In Case Study Research: Principles and Practices (pp. 37 63). New Yo rk: Cambridge University Pess. Gerring, J. (2012). Social Science Methodology: A Unified Framework . Cambridge: University Printing House. Goertz, G. (2006). Structuring and Theorizing Concepts. In Guide . Princeton, NJ: Pri nceton University Press. Goertz, G., & Diehl, P. F. (1993). Enduring Rivalries: Theoretical Constructs and Empirical Patterns. International Studies Quarterly , 37 (2), 147 171. Golub, B., & Jackson, M. O. (2010). Naive Learning in Social Networks and the Wi sdom of Crowds. American Economic Journal: Microeconomics , 2 (1), 112 149.

PAGE 133

124 Goyal, S., & Vega Redondo, F. (2005). Network Formation and Social Coordination. Games and Economic Behavior , 50 , 178 207. http://doi.org/10.1016/j.geb.2004.01.005 Granovetter, M. (1985). Economic Action and Social Structure: The Problem of Embeddedness. American Journal of Sociology , 91 (3), 481 510. Gray, B. (1989). Collaborating: Finding Common Ground for Multi Party Problem s . San Francisco: Jossey Bass. Guiraudon, V. (2000). European Integration and Migration Policy: Vertical Policy making as Venue Shopping. Journal of Common Market Studies , 38 (2), 251 271. http://doi.org/10.1111/1468 5965.00219 Gulati, R. (1999). Network Lo cation and Learning: The Influence of Network Resources and Firm Capabilities on Alliance Formation. Strategic Management Journal , 20 , 397 420. Guo, C., & Acar, M. (2005). Understanding Collaboration Among Nonprofit Organizations: Combining Resource Depend ency, Institutional, and Network Perspectives. Nonprofit and Voluntary Sector Quarterly , 34 (3), 340 361. http://doi.org/10.1177/0899764005275411 Hadden, J., & Jasny, L. (2017). The Power of Peers: How Transnational Advocacy Networks Shape NGO Strategies on Climate Change. British Journal of Political Science , 1 23. http://doi.org/10.1017/S0007123416000582 Hallen, B. L., & Eisenhardt, K. M. (2012). Catalyzing Strategies and Efficient Tie Formation: How Entrepreneurial Firms Obtain Investment Ties. Academy of Management Journal , 55 (1), 35 70. Handcock, M. S., Butts, C. T., Goodreau, S. M., Krivitsky, P. N., & Morris, M. (2017). Fit, Simulate and Diagnose Exponential Family Models for Networks. CRAN. Harder, A., & Gilbert, D. (2015, March 20). Fracking Rules Un veiled by Obama Administration. Wall Street Journal . Heckathorn, D. D. (1997). Respondent Driven Sampling: A New Approach to the Study of Hidden Populations. Social Problems , 44 (2), 174 199. http://doi.org/10.1525/sp.1997.44.2.03x0221m Heckathorn, D. D. (2 011). Snowball Versus Respondent Driven Sampling. Sociological Methodology , 41 (1), 355 366. http://doi.org/10.1111/j.1467 9531.2011.01244.x Heikkila, T., Pierce, J. J., Gallaher, S., Kagan, J., Crow, D. A., & Weible, C. M. (2014). Understanding a Period of Policy Change: The Case of Hydraulic Fracturing Disclosure Policy in Colorado. Review of Policy Research , 31 (2), 65 87. http://doi.org/10.1111/ropr.12058 Heikkila, T., & Weible, C. M. (2015). A Summary Report of a 2015 Survey of the Politics of Oil and Gas Development Using Hydraulic Fracturing in Colorado Authors . Denver. Heikkila, T., & Weible, C. M. (2017). Unpacking the Intensity of Policy Conflict: A Study of Policy Sciences , (March). http://doi.org/10.1007/s11077 0 17 9285 1

PAGE 134

125 Heikkila, T., Weible, C. M., & Olofsson, K. (2017). Lessons From State Level and National Level Policy Conflicts Over U.S. Shale Development. Environment: Science and Policy for Sustainable Development , 59 (3), 4 13. http://doi.org/10.1080/00139157.2017.130 1165 Heikkila, T., Weible, C. M., & Pierce, J. J. (2014). Exploring the Policy Narratives and Politics of Hydraulic Fracturing in New York. The Science of Stories: Applications of the Narrative Policy Framework in Public Policy Analysis , 187 205. 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 . Denver. Henry, A. D., Ingold, K., Nohrstedt, D., & Weible, C. M. (2014a). Policy Change in Comparative Contexts: Applying the Advocacy Coalition Framework Outside of Western Europe and North America. Journal of Comparative Policy Analysis: Research and Practice , 16 (4), 299 312. http://doi.org/10.1080/13876988.2014.941200 Henry, A. D., Ingold , K., Nohrstedt, D., & Weible, C. M. (2014b). Policy Change in Comparative Contexts: Applying the Advocacy Coalition Framework Outside of Western Europe and North America. Journal of Comparative Policy Analysis: Research and Practice , 16 (4), 299 312. Henry , A. D., Lubell, M., & McCoy, M. (2011). Belief Systems and Social Capital as Drivers of Policy Network Structure: The Case of California Regional Planning. Journal of Public Administration Research and Theory , 21 (3), 419 444. Hessels, J., & Terjesen, S. ( 2010). Resource dependency and institutional theory perspectives on direct and indirect export choices. Small Business Economics , 34 (2), 203 220. http://doi.org/10.1007/s11187 008 9156 4 Hill, C. J., & Lynn, L. E. (2005). Is Hierarchical Governance in Decl ine? Evidence from Empirical Research. Journal of Public Administration Research and Theory , 15 (2), 173 195. http://doi.org/10.1093/jopart/mui011 Hillman, A. J., & Hitt, M. A. (1999). Corporate Political Strategy Formulation: A Model of Approach, Participa tion, and Strategy Decisions. The Academy of Mangement Review , 24 (4), 825 842. Hoberg, G., & Phillips, J. (2011). Playing Defence: Early Responses to Conflict Expansion in the Oil Sands Policy Subsystem. Canadian Journal of Political Science , 44 (3), 507 52 7. http://doi.org/10.1017/S0008423911000473 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. http://doi.org/10.1177/01 60323X11428620 Howard, R. M. (2007). Controlling Forum Choice and Controlling Policy: Congress, Courts and the IRS. Policy Studies Journal , 35 (1), 109 123. http://doi.org/10.1111/j.1541 0072.2007.00209.x Hubbert, M. K., & Willis, D. G. (1957). Mechanics of Hydraulic Fracturing. In Transactions of Society of Petroleum Engineers of AIME (pp. 239 257).

PAGE 135

126 Inglehart, R., & Welzel, C. (2005). Modernization, Cultural Change, and Democracy: The Human Development Sequence . Cambridge: Cambridge University Press. Ingold, K. (2011). Network Structures within Policy Processes: Coalitions, Power, and Brokerage in Swiss Climate Policy. Policy Studies Journal , 39 (3), 435 459. http://doi.org/10.1111/j.1541 0072.2011.00416.x Ingold, K., & Leifeld, P. (2016). Structural an d Institutional Determinants of Influence Reputation: A Comparison of Collaborative and Adversarial Policy Networks in Decision Making and Implementation. Journal of Public Administration Research and Theory , 26 (1), 1 18. http://doi.org/10.1093/jopart/muu0 43 Ingold, K., & Pflieger, G. (2016). Two Levels, Two Strategies: Explaining the Gap Between Swiss National and International Responses Toward Climate Change. European Policy Analysis , 2 (1), 20 38. http://doi.org/10.1002/ejoc.201200111 Ingold, K., & Varone , F. (2012). Treating Policy Brokers Seriously: Evidence from the Climate Policy. Journal of Public Administration Research and Theory , 22 (2), 319 346. http://doi.org/10.1093/jopart/mur035 Irvin, R. A., & Stansbury, J. (2004a). Citizen Participation in Dec ision Making: Is It Worth the Effort? Public Administration Review , 64 (1), 55 65. Irvin, R. A., & Stansbury, J. (2004b). Citizen Participation in Decision Making: Is It Worth the Effort? Public Administration Review , 64 (1), 55 65. http://doi.org/10.1111/j. 1540 6210.2004.00346.x Jackson, M. O. (2005). Allocation Rules for Network Games. Games and Economic Behavior , 51 , 128 154. http://doi.org/10.1016/j.geb.2004.04.009 Jackson, M. O., & Watts, A. (2002). On the Formation of Interaction Networks in Social Coor dination Games. Games and Economic Behavior , 41 , 265 291. Jackson, M. O., & Wolinsky, A. (1996). A Strategic Model of Social and Economic Networks. Journal of Economic Theory , 71 , 44 74. Jaffe, M. (2014, September 8). Hickenlooper Names Task Force to Diffu se Drilling Land Use Conflicts. The Denver Post . Jaspal, R., & Nerlich, B. (2014). Fracking in the UK Press: Threat Dynamics in an Unfolding Debate. Public Understanding of Science , 23 (3), 348 363. http://doi.org/10.1177/0963662513498835 Jenkins Smith, H. C. (1990). Democratic Politics and Policy Analysis . (Brooks/Cole, Ed.). Pacific Grove. Jenkins Smith, H. C., Nohrstedt, D., Weible, C. M., & Sabatier, P. A. (2014). The Advocacy Coalition Framework: Foundations, Evolution, and Ongoing Research. In P. A. Sa batier & C. M. Weible (Eds.), Theories of the Policy Process (3rd ed., pp. 183 223). Boulder: Westview Press.

PAGE 136

127 Jing, Y., & Besharov, D. J. (2014). Collaboration Among Government, Market, and Society: Forging Partnerships and Encouraging Competition. Journal of Policy Analysis and Management , 33 (4), 1036 1043. http://doi.org/10.1002/pam Johari, R., Mannor, S., & Tsi tsiklis, J. N. (2006). A Contract Based Model for Directed Network Formation. Games and Economic Behavior , 56 , 201 224. http://doi.org/10.1016/j.geb.2005.08.010 Jones, B. D. (2001). Politics and the Architecture of Choice: Bounded Rationality and Governanc e . Chicago: The University of Chicago Press. Jones, B. D., & Baumgartner, F. R. (2012). From There to Here: Punctuated Equilibrium to the General Punctuation Thesis to a Theory of Government Information Processing. Policy Studies Journal , 40 (1), 1 20. http ://doi.org/10.1111/j.1541 0072.2011.00431.x Jones, M. D., & McBeth, M. K. (2010). A Narrative Policy Framework: Clear Enough to Be Wrong? Policy Studies Journal , 38 (January), 329 353. Kaase, M. (2009). Perspectives on Political Participation. In The Oxford Handbook of Political Behavior (pp. 1 16). Oxford: Oxford University Press. http://doi.org/10.1093/oxfordhb/9780199270125.003.0042 Kaiser, H. F. (1958). The Varimax Criterion for Analytic Rotation in Factor Analysis. Psychometrika , 23 (3), 187 200. Kaiser, H. F. (1974). Kaiser Meyer Olkin Measure for Identity Correlation Matrix. Journal of the Royal Statistical Society , 52 , 296 298. Critical Review: A Journal of Politics and Society , 24 (4), 54 5 567. http://doi.org/10.1080/08913811.2012.788281 Karch, A. (2009). Venue Shopping, Policy Feedback, and American Preschool Education. Journal of Policy History , 21 (1), 38. http://doi.org/10.1017/S0898030609090022 Kingdon, J. W. (1984). Agendas, Alternati ves, and Public Policies . Pearson. Krivitsky, P. N. (2012). Exponential Family Random Graph Models for Valued Networks. Electronic Journal of Statistics , 6 , 1100 1142. Krivitsky, P. N., & Butts, C. T. (2018). Modeling Valued Networks with Statnet. Lee, Y., Lee, I. W., & Feiock, R. C. (2012). Interorganizational Collaboration Networks in Economic Development Policy: An Exponential Random Graph Model Analysis. Policy Studies Journal , 40 (3), 547 573. http://doi.org/10.1111/j.1541 0072.2012.00464.x Leif eld, P. (2013). Reconceptualizing Major Policy Change in the Advocacy Coalition Framework: A Discourse Network Analysis of German Pension Politics. Policy Studies Journal , 41 (1), 169 198. http://doi.org/10.1111/psj.12007 Leifeld, P., & Schneider, V. (2012) . Information Exchange in Policy Networks. American Journal of Political Science , 56 (3), 731 744. http://doi.org/10.1111/j.1540 5907.2011.00580.x

PAGE 137

128 Lewis, K., Gonzalez, M., & Kaufman, J. (2011). Social Selection and Peer Influence in an Online Network. PNAS , 109 (1), 68 72. http://doi.org/10.1073/pnas.1109739109 Ley, A. J. (2014). The Costs and Benefits of American Policy Making Venues. Law and Society Review , 48 (1), 91 126. Lienert, J., Schnetzer, F., & Ingold, K. (2013). Stakeholder Analysis Combined with So cial Network Analysis Provides Fine Grained Insights into Water Infrastructure Planning Processes. Journal of Environmental Management , 125 , 134 148. http://doi.org/10.1016/j.jenvman.2013.03.052 Lowi, T. J. (1972). Four Systems of Policy, Politics, and Cho ice. Public Administration Review , 32 (4), 298 310. Lubell, M. (2004). Collaborative Environmental Institutions: All talk and No Action? Journal of Policy Analysis and Management , 23 (3), 549 573. http://doi.org/10.1002/pam.20026 Lubell, M., Schneider, M., S cholz, J. T., & Mete, M. (2002). Watershed Partnerships and the Emergence of Collective Action Institutions. American Journal of Political Science , 46 (1), 148 163. Lupia, A., & McCubbins, M. D. (1998). The Democratic Dilemma: Can Citizens Learn What They N eed To Know? Cambridge: Cambridge University Press. Lynn, L. E. J., Heinrich, C. J., & Hill, C. J. (2001). Improving Governance: A New Logic for Empirical Research . Washington, D.C.: Georgetown University Press. Malang, T., Brandenberger, L., & Leifeld, P. (2017). Networks and Social Influence in European Legislative Politics. British Journal of Political Science , (grant 149410), 1 24. http://doi.org/10.1017/S0007123417000217 March, J. G., & Olsen, J. P. (1984). The New Institutionalism: Organizational Fact ors in Political Life. American Political Science Review , 78 (3), 734 749. Matti, S., & Sandström, A. (2013). The Defining Elements of Advocacy Coalitions: Coordination and Coalition Structures. Review of Policy Research , 30 (2), 240 257. McConnell, A. (2017). Hidden Agendas: Shining a Light on the Dark Side of Public Policy. Jour nal of European Public Policy , 1 20. http://doi.org/10.1080/13501763.2017.1382555 Mcpherson, M., Smith Lovin, L., & Cook, J. M. (2001). Birds of a Feather: Homophily in Social Networks. Annual Review of Sociology , 27 , 415 444. Mettler, S. (2002). Bringing the State Back in to Civic Engagement: Policy Feedback Effects of the G.I. Bill for World War II Veterans. The American Political Science Review , 96 (2), 351 365. Metz, F., Leifeld, P., & Ingold, K. (2018). Interdependent Policy Instrument Preferences: A Tw o Mode Network Approach. Journal of Public Policy , 1 28. http://doi.org/10.1017/S0143814X18000181 Milbrath, L. W. (1965). Political Participation: How and Why Do People Get Involved in Politics? Chicago, IL: Rand McNally & Company.

PAGE 138

129 Miller Millesen, J. L. ( 2003). Understanding the Behavior of Nonprofit Boards of Directors: A Theory Based Approach. Nonprofit and Voluntary Sector Quarterly , 32 (4), 521 547. http://doi.org/10.1177/0899764003257463 Milward, H. B., & Provan, K. G. (2000). Governing the Hollow Stat e. Journal of Public Administration Research and Theory , 10 , 359 379. Moynihan, D. P., & Soss, J. (2014). Policy Feedback and the Politics of Administration. Public Administration Review , 74 (3), 320 332. http://doi.org/10.1111/puar.12200.Th Mutz, D. C. (20 02). The Consequences of Cross Cutting Networks for Political Participation. American Journal of Political Science , 46 (4), 838 855. Mutz, D. C. (2009). Political Psychology and Choice. In R. J. Dalton & H. D. Klingemann (Eds.), The Oxford Handbook of Polit ical Behavior (p. 10). Oxford: Oxford University Press. http://doi.org/10.1093/oxfordhb/9780199270125.003.0005 Nicholson Crotty, J. (2009). The Stages and Strategies of Advocacy Among Nonprofit Reproductive Health Providers. Nonprofit and Voluntary Sector Quarterly , 38 (6), 1044 1053. http://doi.org/10.1177/0899764009332467 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). Nohrstedt, D. (2013). Advocacy Coalitions in Crisis Resolution: Understanding Policy Dispute in the European Volcanic Ash Cloud Crisis. Public Administration , 91 (4), 964 979. http://doi.org/10.1111/padm.12003 Nohrstedt, D., & Bodin, O. (2019). Collect ive Action Problem Characteristics and Partner Uncertainty as Drivers of Social Tie Formation in Collaborative Networks. Policy Studies Journal , 0 (0), 1 43. Nohrstedt, D., & Olofsson, K. L. (2016a). A Review of Applications of the Advocacy Coalition Framew ork in Swedish Policy Processes. European Policy Analysis , 2 (2), 18 42. http://doi.org/10.18278/epa.2.2.3 Nohrstedt, D., & Olofsson, K. L. (2016b). Advocacy Politics and Strategies on Hydraulic Fracturing in Sweden. In C. M. Weible, T. Heikkila, K. Ingold, & M. Fischer (Eds.), Policy Debates on Hydraulic Fracturing (pp. 147 175). London: Palgrave Macmillan. Nowlin, M. C. (2011). Theories of the Policy Process: State of the Research and Emerging Trends. Policy Studies Journal , 39 (2007), 41 60. Nulty, D. D. ( 2008). The Adequacy of Response Rates to Online and Paper Surveys: What Can Be Done? Assessment & Evaluation in Higher Education , 33 (3), 301 314. http://doi.org/10.1080/02602930701293231 mpact Assessment: Purposes, Implications, and Lessons for Public Policy Making. Environmental Impact Assessment Review , 30 (1), 19 27. http://doi.org/10.1016/j.eiar.2009.05.001

PAGE 139

130 Ba sed Agendas in Public Administration. Public Administration Reviewministration , 57 (1), 45 52. Quest of a More General Theory. Journal of Public Administration Researc h and Theory , 25 (1), 237 256. http://doi.org/10.1093/jopart/muu011 T. (2018). A Dominant Coalition and Policy Change: An Analysis of Shale Oil and Gas Polit ics in India. Journal of Environmental Policy & Planning , 0 (0), 1 16. http://doi.org/10.1080/1523908X.2018.1493984 Olofsson, K. L., Weible, C. M., Heikkila, T., & Martel, J. C. (2017). Using Nonprofit Narratives and News Media Framing to Depict Air Polluti on in Delhi, India. Environmental Communication , 11 , 1 17. http://doi.org/10.1080/17524032.2017.1309442 Olson, M. (1965). The Logic of Collective Action: Public Goods and the Theory of Groups . Cambridge: Harvard University Press. Oser, J., Hooghe, M., & Ma rien, S. (2012). Is Online Participation Distinct from Offline Participation? A Latent Class Analysis of Participation Types and Their Stratification. Political Research Quarterly , 66 (1), 91 101. http://doi.org/10.1177/1065912912436695 Perry, J. L. (1996). Measuring Public Service Motivation: An Assessment of Construct Reliability and Validity. Journal of Public Administration Research and Theory , 6 (1), 5 22. Pfeffer, J., & Salancik, G. R. (1978). The External Control of Organizations . Palo Alto: Stanford University Press. Phillips, W. J. (1972). Hydraulic Fracturing and Mineralization. Journal of the Geological Society , 128 (4), 337 359. Pierson, P. (1996). Increasing Returns, Path Dependence, and the Study of Politics, 94 (2), 251 267. Pralle, S. B. (2003). Venue Shopping, Political Strategy, and Policy Change: The Internationalization of Canadian Forest Advocacy. Journal of Public Policy , 23 (3), 233 260. http://doi.org/10.1017/S0143814X03003118 Pralle, S. B. (2006). Branch ing Out, Digging In: Environmental Advocacy and Agenda Setting . Washington, D.C.: Georgetown University Press. Pralle, S. B. (2009). Agenda Setting and Climate Change. Environmental Politics , 18 (5), 781 799. http://doi.org/10.1080/09644010903157115 Provan, K. G., & Kenis, P. (2008). Modes of Network Governance: Structure, Management, and Effectiveness. Journal of Public Administration Research and Theory , 18 (2), 229 252. http://doi.org/10.1093/jopart/mum015 Provan, K. G., & Milward, H. B. (2001). Do Network s Really Work? A Framework for Evaluating Public Sector Organizational Networks. Public Administration Review , 61 (4), 414 423. http://doi.org/10.5465/APBPP.1999.27628083

PAGE 140

131 Rabe, B. G., & Borick, C. (2013). Conventional Politics for Unconventional Drilling? Lessons Review of Policy Research , 30 (3), 321 340. Rahm, D. (2011). Regulating Hydraulic Fracturing in Shale Gas Plays: The Case of Texas. Energy Policy , 39 (5), 2974 2981. http://doi.org/10 .1016/j.enpol.2011.03.009 Real Dato, J. (2009). Mechanisms of Policy Change: A Proposal for a Synthetic Explanatory Framework. Journal of Comparative Policy Analysis: Research and Practice , 11 (1), 117 143. http://doi.org/10.1080/13876980802648268 Reckhow, S., & Snyder, J. W. (2014). The Expanding Role of Philanthropy in Education Politics. Educational Researcher , 43 (4), 186 195. http://doi.org/10.3102/0013189X14536607 Riccucci, N. M. (2010). Public Administration: Traditions of Inquiry and Philosop hies of Knowledge . Georgetown: Georgetown University Press. Roberts, N. C., & King, P. J. (1991). Policy Entrepreneurs: Their Activity Structure and Function in the Policy Process. Journal of Public Administration Research and Theory , 1 (2), 147 175. Rogers , T., Howard Pitney, B., Feighery, E. C., Altman, D. G., Endres, J. M., & Roeseler, A. G. (1993). Characteristics and Participant Perceptions of Tobacco Control Coalitions in California. Health Education Research, Theory & Practice , 8 , 345 357. Rummel, R. J. (1970). Understanding Factor Analysis. Conflict Resolution , 6 (4), 1 46. Rydin, Y., & Pennington, M. (2011). Public Participation and Local Environmental Planning: The Collective Action Problem and the Potential of Social Capital. Local Environment: The International Journal of Justice and Sustainability , 5 (2), 153 169. http://doi.org/10.1080/13549830050009328 Sabatier, P. A. (1998). The Advocacy Coalition Framework: Revisions and Relevance for Europe. Journal of European Public Policy , 5 (1), 98 130. http ://doi.org/10.1080/13501768880000051 Sabatier, P. A., & Jenkins Smith, H. C. (1993). Policy Change and Learning: An Advocacy Coalition Approach . Boulder, CO: Westview Press. Sabatier, P. A., & Weible, C. M. (2007). The Advocacy Coalition Framework: Innovat ions and Clarifications. In P. A. Sabatier (Ed.), Theories of the Policy Process (Second, pp. 189 220). Boulder, CO: Westview Press. Journal of Extension . Scharpf, F . W. (1989). Decision Rules, Decision Styles and Policy Choices. Journal of Theoretical Politics , 1 (2), 149 176. Schattschneider, E. E. (1960). America . Boston, MA: Wadsworth Cengage Learning.

PAGE 141

132 Schl ager, E. (1995). Policy Making and Collective Action: Defining Coalitions within the Advocacy Coalition Framework. Policy Sciences , 28 (3), 243 270. Schneider, A., & Ingram, H. (1993). Social Construction of Target Populations: Implications for Politics and Policy. American Political Science Review , 87 (2), 334 347. http://doi.org/10.2307/2939044 Schneider, M., Scholz, J., Lubell, M., Mindruta, D., & Edwardsen, M. (2003). Building Consensual Institutions: Networks and the National Estuary Program. American Jo urnal of Political Science , 47 (1), 143 158. Scott, T. A. (2016). Analyzing Policy Networks Using Valued Exponential Random Graph Models: Do Government Sponsored Collaborative Groups Enhance Organizational Networks? Policy Studies Journal , 44 (2), 215 244. Scott, T. A., & Thomas, C. W. (2017). Winners and Losers in the Ecology of Games: Network Position, Connectivity, and the Benefits of Collaborative Governance Regimes. Journal of Public Administration Research and Theory , 27 (4), 647 660. http://doi.org/10. 1093/jopart/mux009 Seawright, J., & Gerring, J. (2008). Case Selection Techniques in A Menu of Qualitative and Quantitative Options. Political Research Quarterly , (1975), 294 308. http://doi.org/10.1177/1065912907313077 Sen, A. K. (1971). Choice Functions and Revealed Preference Theory. Review of Economic Studies , 38 (3), 307 318. Shanahan, E. A., Jones, M. D., & McBeth, M. K. (2011). Policy Narratives and Policy Processes. Policy Studies Journal , 39 (3), 535 561. Shepsle, K. A. (2008). Rational Choice Instit utionalism. In R. A. W. Rhodes, S. A. Binder, & B. A. Rockman (Eds.), The Oxford Handbook of Political Institutions (pp. 23 38). New York: Oxford University Press. Singleton, R. A., & Straits, B. C. (2010). Approaches to Social Research (Fifth). Oxford: Ox ford University Press. Smith, L. G. (1984). Public Participation in Policy Making: the State of the Art in Canada. Geoforum , 15 (2), 253 259. http://doi.org/10.1016/0016 7185(84)90036 8 Sniderman, P. M., & Levendusky, M. S. (2007). An Institutional Theory o f Political Choice. The Oxford Handbook of Political Behavior , (February), 437 457. http://doi.org/10.1093/oxfordhb/9780199270125.001.0001 Snijders, C., Matzat, U., & Reips, U. of Internet Science. International Journal of Internet Science , 7 (1), 1 5. Sobeck, J. (2003). Comparing Policy Process Frameworks: What Do They Tell Us About Group Membership and Participation for Policy Development? Administration & Society , 35 (3), 350 374. http://doi.org/10. 1177/0095399703253052

PAGE 142

133 Sowa, J. E. (2009). The Collaboration Decision in Nonprofit Organizations: Views From the Front Line. Nonprofit and Voluntary Sector Quarterly , 38 (6), 1003 1025. Staw, B. M., Sandelands, L. E., & Dutton, J. E. (1984). Threat Rigidity Effects in Organizational Behavior: A Multilevel Analysis. Administrative Science Quarterly , 26 (4), 501 524. Steglich, C., Snijders, T. A. B., & Pearson, M. (2010). Dynamic Networks an d Behavior: Separating Selection from Influence. Socological Methodology , 40 (1), 329 393. Oil and Gas. New York Times . Thelen, K. (1999). Historical Institutiona lism in Comparative Politics. Annual Review of Political Science , 2 , 369 404. http://doi.org/10.1146/annurev.polisci.2.1.369 Topp, N. W., & Pawloski, B. (2002). Online Data Collection. Journal of Science Education and Technology , 11 (2), 173 178. http://doi .org/10.2307/40188721 Van de Ven, A. (2007). Engaged Scholarship: A Guide for Organizational and Social Research . Oxford: Oxford University Press. Verba, S., & Nie, N. H. (1987). Participation in America: Political Democracy and Social Equality . Chicago, I L: The University of Chicago Press. Watts, A. (2001). A Dynamic Model of Network Formation. Games and Economic Behavior , 34 , 331 341. Webler, T., Tuler, S., & Krueger, R. (2001). What Is a Good Public Participation Process? Five Perspectives from the Publi c. Environmental Management , 27 (3), 435 450. http://doi.org/10.1007/s002670010160 Weible, C. M. (2005). Beliefs and Perceived Influence in a Natural Resource Conflict: An Advocacy Coalition Approach to Policy Networks. Political Research Quarterly , 58 (3), 461 475. http://doi.org/10.1177/106591290505800308 Weible, C. M. (2006). An Advocacy Coalition Framework Approach to Stakeholder Analysis: Understanding the Political Context of California Marine Protected Area Policy. Journal of Public Administration Rese arch and Theory , 17 (1), 95 117. http://doi.org/10.1093/jopart/muj015 Weible, C. M., & Carter, D. P. (2017). Advancing Policy Process Research at Its Overlap with Public Management Scholarship and Nonprofit and Voluntary Action Studies. Policy Studies Journ al , 45 (1), 22 49. http://doi.org/10.1111/psj.12194 Weible, C. M., Elgin, D. J., & Pattison, A. (2011). Political Analytical Capacity and Policy Activities. In Association for Public Policy Analysis and Management 33rd Annual Research Conference (pp. 1 30). Weible, C. M., & Heikkila, T. (2017). Policy Conflict Framework. Policy Sciences , 50 (1), 23 40. http://doi.org/10.1007/s11077 017 9280 6 Weible, C. M., & Ingold, K. (2018). Why Advocacy Coalitions Matter and Practical Insights About Them. Poli cy & Politics , 46 (2), 325 343.

PAGE 143

134 Weible, C. M., Olofsson, K. L., Costie, D. P., Katz, J. M., & Heikkila, T. (2016). Enhancing Precision and Clarity in the Study of Policy Narratives: An Analysis of Climate and Air Issues in Delhi, India. Review of Policy Research , 33 (4), 420 441. http://d oi.org/10.1111/ropr.12181 Weible, C. M., & Sabatier, P. A. (2005). Comparing Policy Networks: Marine Protected Areas in California. Policy Studies Journal , 33 (2), 181 202. Weible, C. M., & Sabatier, P. A. (2006). A Guide to the Advocacy Coalition Framework . In Handbook of Public Policy (pp. 123 1SP United Kingdom: SAGE Publications Ltd. http://doi.org/10.4135/9781848608054 Weible, C. M., Sabatier, P. A., & McQueen, K. (2009). Themes and Variations: Taking Sto ck of the Advocacy Coalition Framework. Policy Studies Journal , 37 (1), 121 140. Weick, K. E. (1995). Sensemaking In Organizations . Thousand Oaks: SAGE Publications Ltd. Wilson, J. Q. (1973). Political Organizations . New York: Basic Books. Wood, Donna, J., & Gray, B. (1991). Toward a Comprehensive Theory of Collaboration. Journal of Applied Behavioral Science , 27 (2), 139 162. http://doi.org/10.1177/0021886391272001 Wu, W. Y., Shih, H. A., & Chan, H. C. (2009). The Analytic Network Process for Partner Selecti on Criteria in Strategic Alliances. Expert Systems With Applications , 36 , 4646 4653. http://doi.org/10.1016/j.eswa.2008.06.049 Zafonte, M., & Sabatier, P. A. (1998). Shared Beliefs and Imposed Interdependencies as Determinants of Ally Networks in Overlappi ng Subsystems. Journal of Theoretical Politics , 10 (4), 473 505. Zaller, J. R. (1992). The Nature and Origin of Mass Opinion . Cambridge: Cambridge University Press.

PAGE 144

135 A PPENDIX A C ONCEPTS AND O PERATIONALIZATION

PAGE 145

136 APPENDIX B FULL SURVEY Question 1: To what extent do you agree or disagree that the following are potential benefits of shale oil and gas development? Strongly disagree (1) Disagree (2) Neither agree nor disagree (3) Agree (4) Strongly agree (5) National energy security Job creation Increase in government revenue through severance, property, and sales taxes A bridge toward renewable energy sources from the natural gas produced Fuel switching from coal to natural gas Reduction of energy costs Decrease in greenhouse gases Other Question 2: To what extent do you agree or disagree that the following are potential problems related to shale oil and gas development? Strongly disagree (1) Disagree (2) Neither agree nor disagree (3) Agree (4) Strongly agree (5) Insufficient capacity by state agencies for regulation Boom and bust economic cycles from natural gas development Contamination of ground and surface water supplies Degradation of air quality Nuisance to the general public caused by truck traffic, noise, and light from well operations Competition over available water supplies Increase in greenhouse gases

PAGE 146

137 Public health impacts from exposure to drilling operations Other Question 3: Please indicate what comes closest to your current position in relation to shale oil and gas development. It should be... Stopped (1) Limited (2) Continued at current rate (3) Expanded moderately (4) Expanded extensively (5) Question 4: If you were to choose between no regulation or one level of government to regulate the following issues related to shale oil and gas development, which would you choose? No regulation (1) Municipal Government (2) County Government (3) State Government (4) Federal Government (5) Water quality Air emissions Disclosure of chemicals in hydraulic fracturing fluids Setbacks of wells from occupied buildings or natural features Location of the wellhead Reclamation o f old well sites Responding to accidents at the well site Water supply Disposing or treating produced water Mitigating public nuisances caused by truck traffic, noise, and light from well site operations Safety of well operators at the well site Question 5: Many political issues in a democracy can be characterized as contentious. Compared to other political issues in Colorado, the level of political contention about shale oil and gas development in Colorado Far less contentious ( 2) Less cont entious ( 1)

PAGE 147

138 Just as contentious (0) More contentious (1) Far more contentious (2) Question 6: Do the views and actions of those you disagree with on shale oil and gas development... Not at all (1) Very little (2) Somewhat (3) Quite a bit (4) A great deal (5) Threaten you personally or professionally (e.g., your job, values, income, or quality of life)? Threaten the state of Colorado? Question 7: Please indicate the extent that you agree or disagree with the following statements. I would support government decisions that would significantly EXPAND shale oil and gas development in Colorado Strongly disagree (1) Disagree (2) Neither agree nor disagree (3) Agree (4) Strongly agree (5) Convincing scientific evidence shows it is completely safe to the environment or public health Convincing scientific evidence shows it boosts the economy Colorado regulators passed and enforced stricter regulations The state provides more authority to local government A majority of Coloradans support its expansion Colorado adopted an energy plan that included a transition away from all fossil fuels Question 8: Please indicate the extent that you agree or disagree with the following statements. I would support government decision s that would LIMIT or STOP Strongly disagree (1) Disagree (2) Neither agree nor disagree (3)

PAGE 148

139 Agree (4) Strongly agree (5) Convincing scientific evidence shows it is a significant threat to the environment or public health Convincing scientific evidence shows it hurts the economy A majority of Coloradans support a ban Mineral right owners were compensated for their potential lost income A catastrophic disaster or emergency occurred from oil and gas development using hydraulic fracturing Colorado significantly expanded its renewable energy production Question 9: To what extent are the interactions with the following groups important in achieving your personal or professional goals related to shale oil and gas development? Not at all important (1) Slightly important (2) Moderately important (3) Very important (4) Extremely important (5) Federal government State government County government City government Oil and gas industry Oil and gas professional associations Environmental or conservation groups Real estate developers or home builders Agricultural organization or farmers Organized citizen groups Churches or other religious organizations Universities or colleges Consulting firms or think tanks Informal personal networks News media Other Question 10: Since I became involved or aware of shale oil and gas development...

PAGE 149

140 I have become more convinced about the benefits (1) My views of the benefits have not changed (2) I have become less convinced of the benefits (3) Question 11: Since I became involved or aware of shale oil and gas development... I have become more concerned about the risks ( 1) My views of the risks have not changed (0) I have become less concerned about the risks (1) Question 12: How would you describe your working professional relationship with people you disagree with on the issue of shale oil and gas development in Colorado? Not collegial at all (1) Somewhat collegial (2) Very co llegial (3) Completely collegial (4) Question 13: How would you describe your working professional relationship with people you agree with on the issue of shale oil and gas development in Colorado? Not collegial at all (1) Somewhat collegial (2) Very col legial (3) Completely collegial (4) Question 14: Are there any organizations or individuals who have the authority and trust to help negotiate policy solutions to shale oil and gas issues in Colorado? Yes (1) No (2) If yes, please indicate the names of a ny such organizations or individuals.

PAGE 150

141 Question 15: To what extent do you think the following ways to influence government are viable for addressing your personal or professional goals for shale oil and gas development? Not viable at all (1) Slightly vi able (2) Moderately viable (3) Very viable (4) Completely viable (5) General elections of government officials Public referendum Regulatory process Legislative process Court/legal process Question 16: Over the past two years, have the following issues in relation to shale oil and gas development in Colorado become worse, stayed the same, or become better? Much worse ( 2) Worse ( 1) About the same (0) Better (1) Much better (2) Government decision making processes Public trust in the COGCC Protection of the environment and public health Economic benefits Greenhouse gas emissions Consideration of vulnerable populations in political decision making Adoption and implementation of effective government regulations Intensity of the political debate Communication by media with the general public about risks and benefits The availability of scientific or technical information Relations between state and local governments Environmental impacts and safety of hydraulic fracturing operations Other

PAGE 151

142 Question 17: Over the past two years, to what extent have you engaged in the following activities and used them effectively in achieving your personal or professional goals related to shale oil and gas development? Not engaged (0) Engaged, but not effective (1) Engaged and moderately effective (3) Engaged and very effective (4) Brokering agreements between parties Countering arguments made by people you disagree with Mobilizing the public Collaborating with people you disagree with Coordinating political activities with allies Providing information to government officials Providing information to the news media Sharing your opinion with government officials Other Question 18: When it Extremely liberal ( 2) Liberal ( 1) Moderate (0) Conservative (1) Extremely conservative (2) Question 19: Please indicate the highest level of education you have attained: Not a High School Graduate (1) High School Graduate (2) Some College (3) Bachelor's Degree (4) Master's or Professional Degree (5) Ph.D. or M.D. (6) J.D. (7)

PAGE 152

143 Question 20: How much of a priority is it for you professionally or personally to deal with political and policy is sues related to shale oil and gas development? Not a priority at all (1) A low priority (2) A moderate priority (3) A high priority (4) The highest priority (5) Question 21: Please indicate your level of experience with the following: No experience (1) V ery little experience (2) Some experience (3) A lot of experience (4) Researching or conducting science on the technical aspects of oil and gas development Reading scientific studies about the economic, environmental, and public health impacts of oil and gas development Analyzing economic or financial impacts of oil and gas development Planning, working, or managing oil and gas operations Owning or leasing mineral or surface rights toward oil and gas development Living within visual proximity of oi l and gas operations Regulating or governing oil and gas development Participating in political activities to influence government decisions about oil and gas development Question 21: What would you recommend, if anything, that might lead to better pro cesses, policies, and outcomes in shale oil and gas development in Colorado?

PAGE 153

144 APPENDIX C FACTOR ANALYSIS (PRINCIPLE COMPONENTS WITH ORTHOGONAL VARIMAX ROTATION)

PAGE 154

145 APPENDIX D BASIC STATISITCAL DESCRIPTION OF THE DATASET VARIABLE N MEAN STANDARD ERROR INDEPENDENT Public 421 Private 421 Nonprofit 421 Anti 353 0.404 0.028 Viability 346 0.640 0.027 Personal Threat 372 2.959 0.075 Colorado Threat 370 3.732 0.063 CONTROLS Education 349 5.123 0.071 Priority 349 3.773 0.044 2015 236