Citation
Examination of policy and political learning

Material Information

Title:
Examination of policy and political learning a study of Colorado climate and energy policy actors
Alternate title:
Study of Colorado climate and energy policy actors
Creator:
Pattison, Andrew ( author )
Language:
English
Physical Description:
1 electronic file (201 pages) : ;

Subjects

Subjects / Keywords:
Political participation -- Colorado ( lcsh )
Climatic changes -- Public opinion ( lcsh )
Environmental policy -- Colorado ( lcsh )
Genre:
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )

Notes

Review:
The central goal of this dissertation is to examine factors that shape policy learning and political learning in policy actors. Policy learning is defined here as change and reinforcement in policy beliefs, and political learning is defined as change and reinforcement of advocacy strategy. The absence of either change or reinforcement was considered “nonlearning.” Factors examined that may affect learning are “extreme” beliefs, and various policy activities. The Advocacy Coalition Framework was used as a theoretical foundation to develop hypotheses. An original survey, with 260 responding climate and energy policy actors in Colorado, provided the data. The results indicate extreme policy beliefs are associated with policy learning and political learning. Further, the product of that learning was more likely to be belief and advocacy strategy reinforcement, as opposed to change. Of the policy activities measured, increased collaboration with policy actors with different views showed some association with mitigation of policy belief reinforcement. Overall, these findings suggest some policy activities can stimulate policy learning and political learning, and may serve as a balance to the reinforcement effects on policy and political learning of extreme beliefs.
Thesis:
Thesis (Ph.D) - University of Colorado Denver.
Bibliography:
Includes bibliographic references
System Details:
System requirements: Adobe Reader.
General Note:
School of Public Affairs
Statement of Responsibility:
by Andrew Pattison.

Record Information

Source Institution:
|University of Colorado Denver
Holding Location:
|Auraria Library
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
944457493 ( OCLC )
ocn944457493
Classification:
LD1193.P86 2015d P37 ( lcc )

Downloads

This item has the following downloads:


Full Text
EXAMINATION OF POLICY AND POLITICAL LEARNING:
A STUDY OF COLORADO CLIMATE AND ENERGY POLICY ACTORS
by
ANDREW PATTISON
B.A., Skidmore College, 1999
M.P.A., University of Colorado, Denver, 2007
A thesis submitted to the
Faculty of the Graduate School of the
University of Colorado in partial fulfillment
of the requirements for the degree of
Doctor of Philosophy
Public Affairs Program
2015


2015
ANDREW PATTISON
ALL RIGHTS RESERVED


ii
This thesis for the Doctor of Philosophy degree by
Andrew Patti son
has been approved for the
Public Affairs Program
by
Paul Teske, Chair
Christopher Weible, Advisor
Tanya Heikkila
Debbi Main
Michele Betsill
Date: October 9, 2015


Ill
Pattison, Andrew (Ph.D., Public Affairs)
Examination of Policy and Political Learning: A Study of Colorado Climate and Energy
Policy Actors
Thesis directed by Associate Professor Christopher Weible
ABSTRACT
The central goal of this dissertation is to examine factors that shape policy
learning and political learning in policy actors. Policy learning is defined here as change
and reinforcement in policy beliefs, and political learning is defined as change and
reinforcement of advocacy strategy. The absence of either change or reinforcement was
considered nonleaming. Factors examined that may affect learning are extreme
beliefs, and various policy activities. The Advocacy Coalition Framework was used as a
theoretical foundation to develop hypotheses. An original survey, with 260 responding
climate and energy policy actors in Colorado, provided the data. The results indicate
extreme policy beliefs are associated with policy learning and political learning. Further,
the product of that learning was more likely to be belief and advocacy strategy
reinforcement, as opposed to change. Of the policy activities measured, increased
collaboration with policy actors with different views showed some association with
mitigation of policy belief reinforcement. Overall, these findings suggest some policy
activities can stimulate policy learning and political learning, and may serve as a balance
to the reinforcement effects on policy and political learning of extreme beliefs.
The form and content of this abstract are approved. I recommend its publication.
Approved: Christopher Weible


IV
DEDICATION
Dedicated to my wife and partner in all adventures, Chandra Russo. Without you
and your support in my life this would, simply put, not have been possible. I do not have
the capacity to thank you appropriately with words, but will spend the rest of my life
trying to do so with actions. I love you. Also to my brothers, and especially my parents,
for giving me a foundation on which to build my life, for inspiring me to always push
myself further, and for always reminding where I came from.


V
ACKNOWLEDGMENTS
There is no way, zero, that I could have done this without the support and
guidance of advisors and mentors in the world of academia. Thank you first to my
dissertation advisor, Chris Weible, for your support and guidance in this process. You
continued to believe in me and push me, long after I deserved your time and energy. Your
dedication to mentoring students, and your pursuit of knowledge, is an inspiration.
Perhaps more than anything, thank you for showing me how brilliant and competent
someone can be, while still being wise enough to be humblea model I will always
aspire to equal, and will likely stumble even coming close to. Thanks man.
Thank you to my dissertation committee. Paul Teske, who walked over to my
cubicle one day in 2006 and told me that I got into the Ph.D. program, that I would be
supported by an NSF IGERT grant, and then explained what that heck that meant. Your
skillful work as the Dean of a fantastic School of Public Affairs, your passion for real
world politics, and most importantly, the way you blend those two in your work in
Denver was the reason I decided to start this process in the first place. Tanya Heikkila,
who was willing to join my committee when I needed her, and provided invaluable
insight and inspiration helping me to think deeply and write simply. Your
professionalism, energy, and humor came to SPA at the perfect time for me, and you
made me believe again that this was something I could, and wanted, to do. Debbi Main,
your passion for community-based participatory research helped me see that our work in
Sustainable Urban Infrastructure can do so much more than improve theories, it can also
improve lives. Michele Betsill, I cited your book on cities and climate change in my
application to the Ph.D. program, and your CV read like how I wanted mine to. I never


VI
dreamt in 2006 that your wisdom and experience would one day be put to improving my
writing, methods, and arguments. Thank you all. This dissertation is better because of
your efforts, and any remaining shortcomings are all mine.
Thank you to other teachers along the way like Linda de Leon, Christine Martell,
Lloyd Burton, and Anu Ramaswami. Thank you especially to George Busenberg and
Peter de Leon, my first two M.P. A. class professors, for opening my eyes. You are all my
inspiration. Thank you to all of my cohort members. Though we have scattered to the
winds over the years, we pushed and pulled each other through those first years with hard
work, and consoled each other with laughs. Thank you especially to Saba Siddiki, John
Calanni, Scott Mendelsberg, and Laurie Mandrino. Thank you to Dallas Elgin for
agreeing to team up on the survey, for advice along the way, and for modeling
professionalism at every point.
I am incredibly grateful to Paul Teske and Anu Ramaswami for the opportunity to
have been supported by the NSF IGERT grant 2007-2010. That sustainable urban
infrastructure program provided incredible financial support and invaluable learning
opportunities. I feel even more blessed and privileged to have been part of that program
with every year that passes. Thank you to Chris Weible, Tanya Heikkila, and Debbi Main
for continuing to lead that program in the later years.
Thank you to my good friends outside the world of UCD SPA. I am blessed to
have too many to name them all here, but you know who you are. For teaching me so
much as we hike, camp, ski, swim, ride, debate, play, laugh, and fish our way through
this fragile and miraculous life. Nothing worth having in life is easy to get, including an
amazing group of friends.


vii
Thank you to my family, and specifically my parents, Art and Kris, for providing
me every opportunity to learn and grow, for having patience with me in the circuitous
path I have taken in all of my employment choices, and for all the support to become my
own person as I did so. Thank you to the Pattison clan, from the first member of the
family to go to college, to a Ph.D. in only two generationsnot too shabby. Thank you to
the Stern clan, with four Ph.Ds. in 14 grandkidswe are so privileged, and so ridiculous.
Finally and most importantly, thank you to Chandra. During the time I worked on
this you were more supportive, empathetic, and understanding than I can believe possible.
You are the rock that weathered all aspects of my insanity and anxiety during this process.
Sorry about that. As always, I look forward to the next adventure with you.


Vlll
TABLE OF CONTENTS
CHAPTER
I. INTRODUCTION......................................................1
Theory: The Advocacy Coalition Framework and Learning.............4
Dissertation Objectives, Research Questions, and Hypotheses
by Chapter.......................................................14
The Case: Colorado Climate and Energy Policy Subsystem...........17
Methods: Data Collection and Analysis............................21
Contributions....................................................23
II. SUMMARY ANALYSIS OF CLIMATE AND ENERGY
POLICY ACTORS IN COLORADO........................................28
Chapter Summary..................................................28
Introduction.....................................................29
The Advocacy Coalition Framework.................................32
Policy Actor Attributes Measured.................................35
Climate and Energy Policy in Colorado............................42
Methodology......................................................44
Results..........................................................48
Conclusions......................................................65
III. FACTORS SHAPING POLICY LEARNING: A STUDY OF
POLICY ACTORS IN SUBNATIONAL CLIMATE AND
ENERGY ISSUES....................................................70
Chapter Summary..................................................70
Introduction.....................................................71
Theory & Hypotheses: Policy Learning and the Advocacy
Coalition Framework..............................................73


ix
Climate Change Policy................................................81
Data and Methods.....................................................84
Results and Analysis.................................................87
Discussion..........................................................104
Conclusion..........................................................108
IV. FACTORS SHAPING POLITICAL LEARNING: A STUDY
OF POLICY ACTORS IN SUBNATIONAL CLIMATE AND
ENERGY ISSUES.......................................................113
Chapter Summary.....................................................113
Introduction........................................................114
Theory and Hypotheses: Political Learning and the Advocacy
Coalition Framework.................................................117
Climate Change Policy...............................................126
Data and Methods....................................................130
Results and Analysis................................................133
Discussion..........................................................140
Conclusion..........................................................145
V. CONCLUSION..........................................................148
Dissertation Summary................................................148
Chapter Two Objectives and Summary Findings.........................149
Chapter Three Research Question, Hypotheses, and Summary
Findings............................................................149
Chapter Four Research Question, Hypotheses, and Summary
Findings............................................................150
Synthesized Findings and Discussion.................................151


X
Limitations......................................................156
Contributions....................................................158
Future Research..................................................166
REFERENCES..............................................................169
APPENDIX
A. Output from Chapter Three Multinomial Logit Regression...181
B. Output from Chapter Four Multinomial Logit Regression............187


XI
LIST OF TABLES
TABLE
2.1. Operational Measures of Attributes...................................45
2.2 Percentage of Respondents That Received Training in Applied
Research............................................................56
2.3. Respondent Information Use Summary...................................59
2.4. Political Learning in Survey Respondents.............................65
3.1. Operational Measures of Attributes...................................86
3.2. Extremeness of Policy Beliefs........................................87
3.3. Policy Learning Across the Six Policy Belief Questions...............89
3.4. Independent Variable Labels in Marginal Change Plots.................93
5.1. Summary of Support for Chapter Three Hypotheses.....................150
5.2. Summary of Support for Chapter Four Hypotheses......................151


xii
LIST OF FIGURES
FIGURE
2.1. Survey Respondents by Political Orientation...............................49
2.2. Organizational Affiliation of Survey Respondents..........................50
2.3. Respondents by Formal Education...........................................51
2.4. Respondents by Years Participating in Climate and Energy Issues...........52
2.5. Respondents by Frequency of Engagement in Policy Activities...............53
2.6. Extent of Policy Capacity in Respondents Organization....................55
2.7. Priority of Climate and Energy-Related Issues.............................56
2.8. Respondents Reported Access to Professional Research......................57
2.9. Policy Beliefs............................................................61
2.10. Policy Learning Across the Six Policy Belief Questions....................64
3.1. Frequency of Policy Activities............................................88
3.2. Changes in Predicted Probabilities in Learning Regarding the
Severity of Predicted Impacts of Climate Change...........................94
3.3. Changes in Predicted Probabilities in Learning Regarding Humans
as the Cause of Climate Change............................................95
3.4. Changes in Predicted Probabilities in Learning Regarding the Need
for Government to Address Climate Change..................................96
3.5. Changes in Predicted Probabilities in Learning Regarding the Need
for a Carbon Tax to Address Climate Change................................97
3.6. Changes in Predicted Probabilities in Learning Regarding the Need
for a Cap & Trade System..................................................98
3.7. Changes in Predicted Probabilities in Learning Regarding the Need
for Policies Promoting Renewables.........................................99
4.1. Policy Actor Extremeness of Belief..........................................134


Xlll
4.2. Learning Regarding Advocacy Strategies......................................135
4.3. Changes in Predicted Probabilities in Learning Regarding
Advocacy Strategy for Climate and Energy Policy Issues......................138


1
CHAPTER I
INTRODUCTION
Nested within the broader field of public affairs is a subset of public policy
research aimed at understanding what has come to be called the policy process. Harold
Lasswell, beginning in the 1940s and 1950s, articulated one of the first theoretical
frameworks used to understand this new field using conceptual maps and a model of
generalizable decision-making processes he called the policy sciences (de Leon, 2006,
p. 39). The modern study of policy process research is defined as the examination of
interactions between public policy and the relevant individuals, events, settings, and
outcomes of policies (Weible, 2014b, p. 5). The study of these interactions can now be
approached from a diverse set of policy process of theories, models, and frameworks
(Cairney & Heikkila, 2014). The Advocacy Coalition Framework (ACF) is one of these
approaches.
The ACF conceives of policymaking as the interaction of competing values of the
opinion leaders and policy implementers working on a particular policy problem. These
specialized individuals, often referred to as policy actors, have invested time and
resources gaining knowledge of specific policy areas and building the relevant political
networks (e.g., advocacy coalitions) to better help affect the development of public policy
(Jenkins-Smith et al., 2014; Kraft & Furlong, 2015; Birkland, 2011). Conflicts can
emerge when there are a diversity of opinions and policy positions held by policy actors
and/or coalitions across a variety of factors. These might include: the severity and
primary cause of environmental problems, the need for the role of the government in
ecosystem protection, and the relative prioritization of specific policy preferences. As


2
policy actors engage and interact, newly acquired information and experiences are
interpreted in ways that may change or reinforce the beliefs and behaviors of the
individual and coalitions engaged in the policy process. These changes in belief and
behavior over time have been defined as learning. The concept of learning has long
played a central role in the ACF and other theories and frameworks used to understand
policy processes (Lindblom, 1956; Heclo, 1974; Bennett & Howlett, 1992; May, 1992;
Sabatier & Jenkins-Smith, 1993; Gerlak & Heikkila, 2013).
This study will differentiate between policy learning, pertaining to cognitive
change and/or reinforcement in policy beliefs, and political learning, regarding behavioral
change and/or reinforcements in advocacy strategy employed in policymaking. Policy
actor learning, defined as belief and behavior change or belief and behavior
reinforcement, may be shaped by factors such as cognitive abilities, perceptual filters, or
existing belief structures (Lord et al., 1979; Simon, 1985). Other factors affecting
learning may be experiences such as participation in policy activities between ally and
opponent coalitions that advocate different policy goals (Sabatier & Jenkins-Smith, 1993,
1999). For example, the extent to which policy actors engage in policy activities such as
collaboration and consensus building with individuals or coalitions with dissimilar policy
views and goals may affect information acquisition, and potentially shape learning
(Sabatier & Jenkins-Smith, 1993, 1999; Jenkins-Smith et al., 2014). This dissertation asks
and seeks to answer the following general question: What factors influence how
information and experiences shape policy actor learning? Some factors may lead to
different forms of learning. For instance, some factors may shape learning as changes in
beliefs and behavior and some factors may reinforce beliefs and behavior. Acquiring a


3
better understanding of these factors may help to build more effective policy processes,
processes better equipped to navigate what might appear as intransigent conflicts between
individuals and groups working on issues such as climate and energy policy.
The field of policy process theory is founded on the observation that public policy
does not develop within a single government institution, but by subunits of a political
system in what have been called policy subsystems (Redford, 1969). Policy subsystems
continue to be a primary unit of analysis in the field of policy process as it has been
evident from decades of research that policies often are created and evolve through the
work of multiple overlapping involved organizations, groups, and coalitions of
specialized policy actors (Birkland, 2011; Sabatier & Jenkins-Smith, 1993). Using a
study of the Colorado climate and energy policy subsystem, and the ACF as a lens
through which to view the attributes, beliefs, and learning of the policy actors therein, the
major objectives of this dissertation are to:
1) Measure and understand the diversity of various attributes of these policy
actors and the extent of policy actor policy and political learning in this subsystem.
2) Examine what, if any, effects extreme beliefs and policy activities have on
policy actor policy learning.
3) Examine what, if any, effects extreme beliefs and policy activities have on
policy actor political learning.
THEORY: THE ADVOCACY COALITION FRAMEWORK AND LEARNING
The Advocacy Coalition Framework (ACF), developed by Sabatier and Jenkins-
Smith in the late 1980s, evolved from a long history of research on environmental
problems, and was an attempt to reconcile top-down and bottom-up theories of policy


4
implementation (Sabatier, 1988, 1998; Jenkins-Smith et al., 2014). Throughout its
development, the ACF has become a leading framework to examine and explain long-
term policy change, individual policy actor belief and behavior, and changes in beliefs
and strategies (Jenkins-Smith et al., 2014). As with other social science theories and
frameworks, the ACF builds from the social-psychology model of the boundedly rational
individual possessing perceptual limits and filters stemming from preexisting belief
structures (Sabatier, 1988; Sabatier & Jenkins-Smith, 1993). The primary unit of analysis
assumed by the ACF is the policy subsystem. According to the ACF, two dimensions
characterize a policy subsystem: a substantive dimension and a territorial dimension
(Jenkins-Smith et al., 2014). In this study, the substantive dimension is climate and
energy policy (specifically in this case regarding policy pertaining to climate change,
greenhouse gas mitigation, or climate action in general), and the territorial dimension is
Colorado (meaning policy actors who are active within the State of Colorado, and
Colorado cities, for example the City and County of Denver).
Research under the ACF has focused on a variety of phenomenon, such as the
extent that policy actors are motivated by and filter newly acquired information through
their beliefs while attempting to influence policymaking. The logic of such a pursuit
comes from an assertion that individual beliefs are the causal driver for political behavior,
and pull like-minded actors together in advocacy coalitions (Sabatier, 1988). The ACF
defines advocacy coalitions as groups of legislators, agency officials, interest group
leaders, and researchers with similar policy beliefs that share resources and engage in a
nontrivial degree of coordination (Jenkins-Smith et al., 2014).


5
Policy actors in the climate and energy policy subsystem are expected to have a
variety of beliefs regarding causal mechanisms, such as the relative contribution to
natural global climate change caused by human behavior and specific policy preferences,
like the need for government intervention into the energy market. Belief and value
clashes within the policy process are likely involved in the lack of substantive national
climate policy, but policy change has been occurring at the local level over time, despite
goal conflicts and technical disputes. Potentially, this indicates learning is occurring in
local energy and climate policy subsystems, thus indicating an appropriate case to use the
ACF to test some hypotheses related to policy actor learning.
Theoretical Emphasis
The ACF puts forth a number of causal relationships, which serve as a theoretical
foundation in this study (for a full list of ACF hypotheses see: Jenkins-Smith et al., 2014).
There are three major ACF theoretical emphases, which applications in the literature
typically focus upon: 1) advocacy coalitions (understanding and explaining coalition
structure, membership, resources, and changes over time); 2) learning or policy-oriented
learning (changes in policy actor beliefs or the use of political strategies for achieving
objectives); and 3) policy change (changes in core aspects of governmental policies or the
beginning or end of specific programs) (Jenkins-Smith et al., 2014). This study will focus
on theories regarding learning and will specifically examine both policy learning
(pertaining to changes and reinforcement in policy beliefs) and political learning
(pertaining to behavioral changes and reinforcements in advocacy strategy).
To contextualize this studys examination of policy and political learning, some
discussion of the way learning is studied in disciplines other than public affairs is necessary.


6
Various social and natural science fields have struggled to understand learning for decades
(Grin & Loeber, 2007; Milner, Squire & Kandel, 1998). Despite a great deal of scholarship
in regards to this topic, it is difficult to make comparisons across much of the work. Muro
and Jeffery (2008, p. 327) claim that this difficulty is due to the ... different underlying
assumptions about the nature of learning of knowledge.... In other words, different
conceptualizations of learning within and across disciplines have confounded attempts to
make synthesizing theories and models of individual learning (Reed et al., 2010). Recent
meta-studies of learning (see Murro & Jerry, 2008; Reed et al., 2010; Crona & Parker,
2012) agree that while no singular definition, conception, or operationalization of learning
across the social sciences exists, learning is often cited as a normative goal, and perhaps a
prerequisite for collaborative and adaptive governance and sustainable policymaking. It is
therefore important to delineate different dimensions of learning.
As in other disciplines, the investigation of learning in the policy process literature
has a long and eclectic history (Heikkila & Gerlak, 2013; Freeman, 2006; Grin & Loeber,
2007). Heclos (1974, pp. 303-306) discussion of learning as collective puzzling as a
factor in policy change is often cited in public policy literature and influenced Sabatiers
early work in developing the ACF. Bennett and Howletts (1992) effort to synthesize
theories of learning in public affairs is useful to unpack some of the various dimensions of
learning and may allow for larger lessons to be drawn from the complicated field of
learning. Bennett and Howlett (1992) describe a framework to describe learning studies
across three dimensions: 1) who is learning, 2) what is being learned, and 3) the results of
learning. For example, individual policy actors may learn about the efficacy of certain
climate and energy policies, leading to policy change over time.


7
In terms of who is learning, distinctions must be made between individual
learning and group learning (Reed et al., 2010). Mixing methodologies or concepts across
different units of analyses can lead to difficulties. This dissertation is examining
individual learning to better understand the factors that shape policy actor learning in this
policy subsystem. In terms of what is being learned, individuals can learn about any
aspects of public policies. May (1992) provides a useful distinction between social policy
learning (the social construction of the policy problem or goal for example), instrumental
policy learning (such as the efficacy of various policy tools or designs), and finally
political learning (which would involve knowledge of the viability or public palatability
of various policy tools or instruments). This dissertation explores both policy learning
and political learning. Policy learning (encompassing both social policy learning and
instrumental policy learning) will be examined in the form of change and reinforcement
of policy actors policy beliefs. Political learning will be measured in the form of policy
actors advocacy strategy change and reinforcement.
The typologies of learning defined by Bennett and Howlett (1992) and May
(1992) helped organize the public affairs literature initially, and recent studies are
attempting to investigate further, and explore the role of learning in the policy process
(Heikkila & Gerlak, 2013; Gerlak & Heikkila, 2011). A crucial differentiation is that the
results of learning (such as belief change or reinforcement of advocacy strategies) are
distinct from the process of learning itself, meaning the factors that shape the learning,
such as various sources of information or policy activities. The results or outcomes of
learning may be conceptualized in numerous ways, for instance as policy change
(Busenberg, 2001). Some researchers make distinctions between cognitive versus


8
behavioral changes (Gerlak & Heikkila, 2011), while others look for organizational
changes in collaborative settings (Crona & Parker, 2012). The contextual factors that
describe how social environments and social networks cultivate learning also seem
important to delineate (Heikkila & Gerlak, 2013; Reed et al., 2010), as do broader
descriptions of the political environment (Gerlak & Heikkila, 2011).
Learning, as a part of the policy process, is particularly relevant within the ACF.
ACF scholars have explicitly argued since its inception that, in the attempts by various
policy actors to affect policies and programs, efforts are made to improve their
understanding of the problem and alternatives, thus leading to individual learning
(Sabatier, 1988)1. Within the ACF, learning has been defined as continuing adjustments
to beliefs and behaviors concerning policy objectives resulting from new experiences or
information (Sabatier & Jenkins-Smith, 1999, p. 123). However, learning has been
inconsistently and/or poorly defined both theoretically and operationally across ACF
applications (Weible et al., 2011). Many applications of ACF have affirmed some of the
underlying theoretical claims. For instance, that opposing advocacy coalitions within
policy subsystems are formed and maintained based on stable policy beliefs, and that
policy change is associated with major policy or political events (Jenkins-Smith et al.,
2014). Learning, however, has been the least explored of the three major ACF theoretical
emphases (Jenkins-Smith et al., 2014; Weible et al., 2009; Zafonte & Sabatier, 1998).
In the studies conducted over the previous decades, scholars applying the ACF
have found learning to be more likely in a context where: 1) conflict is at intermediate
1 In the introduction of the ACF, Sabatier used the term policy-oriented learning. This
study will follow the lead of other scholars employing the ACF and abbreviate this as
learning.


9
levels and focused on beliefs related to specific policy design alternatives, as opposed to
more fundamental beliefs such as causal arguments or political orientation; 2) issues in
conflict are technical in nature (as opposed to social); and 3) when there are professional
forums for individuals in opposing coalitions to collaborate on policy development
(Lester & Hamilton, 1987; Sabatier & Brasher, 1993; Eberg, 1997; Ellison, 1998; Lubell,
2003; Meijerink, 2005; Larsen et al., 2006). A 2009 (Weible et al.) comprehensive meta-
analysis of applications of the ACF found that some studies have shown policy learning
and individual belief change within and across coalitions to have potential to drive policy
development and change over time. But multiple subsequent surveys of ACF theory and
applications argue that one major area within the ACF needing innovation and improved
clarity in theory and model-building, as well as methodology, is learning (Jenkins-Smith
et al., 2014; Weible et al., 2011). By exploring both policy learning and political learning
as distinct concepts, this dissertation aspires to contribute to the improvement of the study
of learning in the policy process.
Using the ACF to Study Learning
One of the essential challenges in studying learning, in general and using the
ACF, is how learning is defined conceptually. For example, is policy actor belief change
the only necessary and sufficient indicator of learning? Or is coalition change, or is
policy change that indicator? In the ACF literature, one or more of these three
indicators of learning are often provided as evidence of learning, but rarely is this
explicitly linked to a specific conceptualization of learning. Even fewer ACF applications
have attempted to measure learning quantitatively using survey data; instead, researchers


10
have typically relied on unsystematic assessments of policy change or content analysis
(Weibleetal., 2011).
Other potential indicators of learning posited by the ACF are belief reinforcement,
advocacy strategy change, and advocacy strategy reinforcementall of which may occur
based on different learning processes of policy actors. For instance, some experiences in
the policy process may serve to change or refine policy actors beliefs and behaviors,
while participation in different policy activities may serve to strengthen and galvanize
beliefs and behaviors (Heikkila & Gerlak, 2013). Relatedly, some attributes, such as
extremeness of preexisting beliefs, may factor into the learning process of policy actors.
To further narrow a conceptual definition of learning, the various dimensions of
the learning concept from Bennett and Howlett (1992) and May (1992) are again helpful.
To begin, in this study, the who is learning will be policy actors in the climate and
energy subsystem of Colorado. Individual policy actors will be the unit of analysis.
As to the what is being learned, this study focuses on two aspects of learning.
Chapter Three examines policy learning (or social and instrumental policy learning using
Mays 1992 typology) by examining cognitive belief change and belief reinforcement, as
well as the absence of either, regarding a range of policy information across the energy
and climate policy subsystem. Specifically, policy beliefs related to causal mechanisms
and the perceived severity of the problem of climate change will be examined.
Additionally, beliefs examined regarding the role the government (versus the economic
market) should play in addressing the problem, as well as beliefs related to the perceived
need for various specific policy solutions. This kind of learning related to the social
construction of the problem, and learning regarding the viability of specific policy


11
implementation designs, are exemplars of social policy learning and instrumental policy
learning respectively in Mays (1992) typology.
Chapter Four explores political learning by measuring behavioral change and
reinforcement of advocacy strategies, as well as the absence of either. Thus, policy
actors beliefs and behaviors are included within the dimension of what is being
learned in this study, and thus span across all three of Mays (1992) typology of policy
and political learning dimensions. Other fields, such as sociology, recognize and study
the role of political actors strategy responses based on changing political opportunities
and threats in policymaking processes, contentious or otherwise (Tilly & Tarrow, 2007).
The ACF postulates that coalition and individual actors advocacy strategies may shift,
based on, for instance, changes in short-term political constraints or resources (Jenkins-
Smith et al., 2014). This changing or reinforcing of advocacy strategy based on
awareness of political feasibility is an example of Mays (1992) political learning. Within
the context of subnational climate and energy policy actors specifically, actors may adjust
advocacy strategies based on political factors regarding the perceived need for local
governments to address climate change or sociotechnical factors such as trends in energy
production and infrastructure construction (Bulkeley, 2013). The direct examination of
policy actors political learning regarding advocacy strategies is underdeveloped
compared to issues related to coalition resources and activities more broadly (Jenkins-
Smith et al., 2014).
Using Bennett and Howlett (1992) typology, this study defines policy actors
reports of their cognitive and behavioral changes (or reinforcements) as the results of
learning. These cognitive and behavioral reinforcement and changes could also be


12
categorized as what Heikkila and Gerlak (2013) would refer to as the products of
learning. This dissertation seeks to specifically differentiate between belief change and
belief reinforcement, as well as change and reinforcement of advocacy strategies, as
different products of policy learning and political learning respectively. While change
and reinforcement (cognitively or behaviorally) may be included in a concept of learning
in the ACF, this study argues they are different categories of learning with different, and
perhaps different, implications. Understanding the factors that shape learning, the
processes of learning resulting in different products, and delineating between change and
reinforcement, is the explicit goal of this study. Chapter Three of this study specifically
conceptualizes and defines belief reinforcement in addition to belief change, as evidence
of conative or policy learning. Similarly, Chapter Four conceptualizes and defines
reinforcement in addition to change, in policy actor advocacy strategies as evidence of
behavioral or political learning.
Another challenge in studying the concept of learning using the ACF, or any set
of policy process theories or frameworks, is understanding what Goertz (2006, p. 178)
would refer to as the negative pole or the negative concept. Specifically, if cognitive
and/or behavioral change, as well as reinforcement, can constitute learning, what
constitutes non-learning? This question is rarely considered in studies of learning
across many fields of study (Heikkila & Gerlak, 2013). Evidence of non-learning could
be, for example, the absence of belief change or policy change. Interpretations of early
applications of policy learning theories of the ACF would seem to indicate that since
policy change is prima fascia evidence of learning, then no policy change is evidence of
non-learning. But that logic rests on the assumption that policy change indicates and


13
comprises the concept of learning, and that may not always be true. In science, when
measuring a phenomenon, it is preferable to measure variables across categories of
positive and negative states. While scale data would be ideal, it is not possible with all
conceptualizations of learning. Within the ACF, learning is defined as: relatively
enduring alterations of thought or behavioral intentions that result from experiences
and/or new information and that are concerned with the attainment or revision of policy
objectives (Sabatier & Jenkins-Smith, 1999, p. 123). In this case, alterations includes
both change and reinforcement in beliefs and behaviors. Thus, absence of change or
reinforcement (in either policy beliefs on in political behaviors) will be evidence for not
learning.
In Chapter Three, policy learning will be measured as self-reported changes or
reinforcement of beliefs across six belief structures. Chapter Four will examine political
learning as self-reported change or reinforcement of advocacy strategies. Individuals are
asked directly if their beliefs and strategies have been changed or reinforced, as well as
what the source of that change or reinforcement was. This dissertation is one of the first
ACF applications to specifically ask policy actors to report learning as change and
reinforcement, measure both, as well as the absence of either, and ask about the source of
that learning. Multinomial logistic regression analyses were used to examine the role of
different factors in shaping different policy learning and political learning products.
This dissertation aims to explore, test, and refine models of individual policy actor
learning, specifically, to examine some of the factors that shape policy learning and
political learning in this policy subsystem. In employing the ACF, all attempts have been


14
made to clarify assumptions, methods, and approach. This dissertation is written as three
stand-alone chapters with an introduction and conclusion. This study begins by providing
an overall descriptive analysis of various attributes of the policy actor survey respondents
including, but not limited to, the distribution of variation in beliefs, behaviors, and
learning. The second and third chapters examine factors shaping policy learning and
political learning respectively within this sample of policy actors.
DISSERTATION OBJECTIVES, RESEARCH QUESTIONS,
AND HYPOTHESES BY CHAPTER
Chapter Two: Summary Analysis of Climate and Energy Policy Actors in Colorado
Chapter two provides an overall analysis of the sample of policy actors, including
demographics, for instance, that responded to the survey. Descriptive statistics are
provided for a host of variables measured in the sample, including the extent of policy
learning and political learning occurring in this subsystem.
The two main objectives guide this chapter to:
1) Understand and describe attributes of the policy actors (individuals in the
government, private, nonprofit, and academic sectors) involved in climate and energy
policy debates in Colorado, including their beliefs across a number of relevant policy
questions, and
2) To determine the extent of policy learning, as defined as the range between
belief change and belief reinforcement, and the extent of political learning, as defined
by a change or reinforcement of advocacy strategies, in the sampled population of
policy actors.


15
Chapter Three: Factors Shaping Policy Learning: A Study
of Policy Actors in Subnational Climate and Energy Issues
Chapter Three examines policy learning, as measured by belief changes and
reinforcement, in the sample set of policy actors that replied to the survey. Various
factors are measured to determine the effect on policy learning.
Research Question:
What factors affect policy actor belief change and reinforcement in a local and
state level energy and climate policy subsystem?
In order to explore the effect of these factors on policy learning, the following
four hypotheses are presented and tested for using survey data.
Extreme Belief Hypothesis:
Hypothesis 1: Policy actors with more extreme policy views are more likely to
reinforce their beliefs than change their beliefs.
Policy Activities Hypotheses:
Hypothesis 2: Policy actors that seek advice more frequently from those with
similar beliefs are more likely to reinforce their beliefs, rather than change their beliefs.
Hypothesis 3: Policy actors that collaborate more frequently with those with
dissimilar beliefs are more likely to change their beliefs, rather than reinforce their beliefs.
Hypothesis 4: Policy actors that have participated in more frequent facilitated
consensus-based processes are more likely to change their beliefs, rather than reinforce
their beliefs.


16
Chapter Four: Factors Shaping Political Learning: A Study
of Policy Actors in Subnational Climate and Energy Issues
Chapter Four examines political learning, as measured by change and
reinforcement of advocacy strategies, of the policy actor respondents to the survey.
Different variables are measured to determine the effect on political learning.
Research Question:
What factors affect policy actor political learning in a local and state level energy
and climate policy subsystem?
With the goal of exploring the effect of these factors on political learning, three
hypotheses were examined using data from the survey respondents.
Extreme Belief Hypothesis:
Hypotheses 1: Policy actors with extreme beliefs will be more likely to change or
reinforce advocacy strategies than not change or reinforce advocacy strategies.
Policy Activities Hypotheses:
Hypothesis 2: Policy actors that participate in coalition building will be more
likely to change or reinforce advocacy strategies than policy actors that do not participate
in coalition building.
Hypothesis 3: Policy actors that participate in facilitated multi-stakeholder
consensus-based processes will be more likely to change or reinforce advocacy strategies
than policy actors that do not participate in negotiated multi-stakeholder consensus-based
processes.
Hypothesis 4: Policy actors that participate in coalition building will be more
likely to report advocacy strategy change than advocacy strategy reinforcement.


17
Hypothesis 5: Policy actors that participate in facilitated multi-stakeholder
consensus-based processes will be more likely to report advocacy strategy change than
advocacy strategy reinforcement.
THE CASE: COLORADO CLIMATE AND ENERGY POLICY SUBSYSTEM
The findings of the Intergovernmental Panel on Climate Change, and the broader
scientific community, demonstrate that climate change could have dramatic impacts on
human communities in the near future (Giddons, 2011; IPCC, 2013, 2014; Stem, 2007).
Despite this threat, the U.S. has yet to produce meaningful federal climate legislation2.
This has provided the opportunity for climate policy innovation at the subnational level.
Cities, towns, counties, and states have created and implemented a litany of climate and
energy programs and legislation in the U.S. Thus, the study of these subnational climate
policy subsystem case studies has become important among political scientists and
scholars struggling to understand the role of policy actors and the policy process.
Previous research, specifically regarding subnational climate policy, has led to
insights regarding the need to further examine the actions of policy makers (Davis &
Weible, 2011), the drivers of local and state-level decisions to commit to climate
protection (Rabe, 2004; Fogel, 2007; Selin & VanDerveer, 2007; Zahran et al., 2008;
Krause, 2011), and issues of local climate governance (Betsill, 2001; Bulkeley & Betsill,
2003; Bulkeley, 2013). This progress in the study of the politics and policies of
subnational units has evolved alongside efforts in the engineering and other physical
sciences. Pioneering work related to the creation of standardized methods for greenhouse
2 At the time of writing, President Obama is attempting to advance more substantive
climate policies at the national level, but political constraints on implementation have
continued to slow progress.


18
gas accounting which was done at the subnational level, and especially city level, focused
on: modeling urban carbon flows with ecosystem theory (Churkina, 2008), urban
demand-centered approaches to differentiate cross-political jurisdictional and point-of-
use emissions (Hillman & Ramaswami, 2010; Ramaswami et al., 2008), and the
relationship between carbon emissions and urban development (Glaeser & Kahn, 2010).
Progress in the development of meta-theories to understand and improve social-
ecological systems management and subnational climate policy, specifically, will be an
interdisciplinary effort (Ramaswami et al., 2012b). This study strives to be one aspect of
that effort in further developing models of individual learning of policy actors and will
employ a single policy subsystem study designed to explore factors that shape learning,
by applying the theoretical lenses of the ACF to the climate and energy policy subsystem
of Colorado.
Subnational Climate Policy Landscape
The Kyoto Protocol, the international agreement to address climate disruption
with changes to energy policies, went into effect in 2005, and almost 200 countries have
ratified it to date. While the U.S. is a signatory to the Kyoto Protocol, Congress has
continually refused to formally ratify the treaty, preventing full U.S. participation (Regan,
2015; Layzer, 2006)3. In an effort to advance the goals of the Kyoto Protocol through
local government leadership and action, on the same day the treaty went into effect across
the globe, then Seattle Mayor Greg Nickels launched the U.S. Mayors Climate Protection
Agreement (USMCPA) (US Conference of Mayors, 2015). Currently, there are more than
3 At the time of writing, the Obama Administration is attempting to negotiate a more
comprehensive international climate treaty. Given the complexities and uncertainties of
international agreements, and the need for Congressional ratification of U.S. treaties, the
details of any eventual policy outcome are currently unknowable.


19
1,000 signatories to the USMCPA, including Denver, Colorado (US Conference of
Mayors, 2015). Participating cities have agreed to reduce community-wide greenhouse
gas (GHG) emissions by 2012 to at least 7 percent below 1990 levels, or better, in their
own communities by using a litany of climate-related polices and plans.
Climate and energy policy at the state level has also been created. Early climate
and energy policy innovations at the state level were categorized and examined by Rabe
(2004). While many of the policies were symbolic at first, later and especially recent
developments have become much more aggressive in terms of the stated goals and the
policy details (Bulkeley, 2013; Krause, 2011; Ramseur, 2007). At least 30 U.S. states,
including Colorado, in addition to hundreds of U.S cities, have now created a climate
action plan (CAP) of some sort (EPA, 2015). A CAP typically outlines policy goals and
recommendations that a state or city will employ to address climate change by making
specific policy actions toward reducing the GHG emissions of that entity.
Denver and Colorado Climate Policy Landscape
Then Mayor John Hickenlooper launched Greenprint Denver in 2005. This new
City and County of Denver department was created to advance and further support the
integration of environmental impact analysis into the citys programs and policies,
alongside economic and social analysis (City and County of Denver, 2006, p. 2). The
program built on the history of Denver climate and energy policy, going back a decade as
an early member of the International Council of Local Environmental Initiatives (now
know simply as ICLEI) (Bulkeley & Betsill, 2003). The Greenprint Denver Plan includes
a 10-point action agenda to reduce the citys environmental impact. It is a comprehensive
list covering a variety of environmental policies. The research study proposed here will


20
concern those action items dedicated to climate change policy and action. The first and
perhaps most ambitious action agenda item from the Greenprint Denver Plan is to reduce
Denver per capita greenhouse gas emissions by 10% below 1990 levels by 2011. Work in
partnership with regional mayors, universities, and the business community to develop and
implement effective strategies for adaptation to and amelioration of global climate change
(City and County of Denver, 2006, p. 6). Compared to other medium to large U.S. cities
that are signatories to the USMCPA, Denver is typical in terms of its population density,
stated climate policy goals, and current progress in attaining said goals.
In November 2007, then Colorado Governor Bill Ritter launched an initiative to
address climate change statewide, which resulted in the creation of the Colorado Climate
Action Plan. The plan called for a reduction of the state emission of greenhouse gases by
20 percent by 2020, and was created in a collaborative manner from a diverse set of
stakeholders ... including business and community leaders, conservationists, scientists
and concerned citizens (Ritter, 2007, p. 2). The Colorado CAP is similar to the
approximately 30 other state plans in the U.S. (EPA, 2015; Ramseur, 2007).
Both the Greenprint plan and the Colorado CAP were preceded by a series of
roundtable discussions and public input sessions that provided both the formal and
informal discussions between advocates and opponents of citywide climate policies.
Since the inception of both plans, Mayor Hickenlooper has become the Governor of
Colorado and is now leading the charge on the state CAP. The policy actors from across
Colorado, including those from Denver, will be treated together as subnational policy
actors for the following reasons: 1) the overlap in city and state political actors;
2) Denver is the largest city in Colorado, the states capital, and is influential in state-


21
level policy developments in climate and energy issues; and 3) often, as in this case, state
and city governments cooperate on energy and climate policy. Because the Denver and
Colorado plans are consistent with the sets of city and state climate action plans in
existence (EPA, 2015; Ramseur, 2007), a typical-case approach to case selection is,
according to Gerring (2007, pp. 91-93), useful for a study such as this.
METHODS: DATA COLLECTION AND ANALYSIS
This study is an investigation of the climate and energy policy subsystem of
Colorado. This study uses the ACF to investigate change and reinforcement of beliefs
and advocacy strategies at the individual level, using data collected from a 2011 original
cross-sectional survey. The unit of analysis, as is consistent with similar applications of
the ACF in the academic literature cited above, is the individual policy actor active in
this subsystem, such as government employees for instance. To insure reliability and
replicability of this study, the research steps will be explicated below, along with
validity issues.
Data Collection
At the first step, a small number of preliminary open-ended interviews were
conducted with key policy actors to learn about the history of the subsystem, refine the
instruments, and introduce this project to the climate change policy community. The
interviewees were asked for feedback on a draft version of the survey, which helped to
determine survey content, especially as it relates to specific policy options and details.
These individuals were not a formal advisory council, but aided in the primary goal of
insuring the final surveys would be accurate, relevant, and have good face validity in
terms of variable measurement.


22
The next step in the data collection process was to formalize the target population
for the survey. As is indicated and consistent with the literature, policy actors of the
climate and energy policy subsystem in Colorado were the target population of this study.
Policy actors (sometimes referred to as policy participants or policy elites) are defined as
individuals involved in the policy subsystem. The individuals were identified as being
employed by, or involved with, government agencies, non-governmental organizations,
and private companies participating in climate policymaking in Colorado. Three
strategies were used in generating a list of study participants. The first group targeted for
inclusion in the study was the individuals who attended the roundtable sessions preceding
the creation of state and local government policies, and those who attended relevant
public input sessions during the policymaking process. Second, as is consistent with
applications of the ACF, a snowball sample was generated from the lists, with input from
the preliminary interviewees. Third, the major governmental and non-governmental
organizations in Colorado involved in climate and energy policy were identified and the
directors, board members, and key staff or officials were also targeted for inclusion.
These sampling strategies have been used successfully in much of the literature cited
above in applying ACF theories to policy subsystems.
The completion of the survey was step two of the data collection process. As
suggested by Fowler (2002) and mentioned above, the survey was beta-tested on the
informal preliminary interviewees to confirm face validity and insure clarity and
comprehension. Knowing the expected response rate of the final questionnaire could be
approximately 30-60 percent, the range of response rates of other studies employing
similar techniques (Fowler, 2002). Surveys were emailed to enough policy actors to attain


23
at least a 30 percent response rate, a goal set to capture adequate variation in the sample
population. In order to increase the response rate, the survey was designed to take only
approximately 20-30 minutes to complete, and timely reminders were sent. In the end,
the survey was sent to 793 policy actors involved in climate and energy issues in
Colorado, and 260 individuals returned fully completed surveys (response rate of 33
percent). The sampling technique that was employed for the survey is explicitly
purposive and non-probability based, as is consistent with the above-cited literature. The
reason for the use of this technique is that the views and opinion of active policy actors is
explicitly the goal of this study.
Data Analysis
The quantitative data collected with the surveys were analyzed using STATA13
software. Descriptive statistics and crosstabs were run summarizing policy actor attributes
and learning among the sample, and are presented in Chapter Two. Multinominal logistic
regression (MNLR) analysis was used to test the effects, extremeness of beliefs, policy
activities, and sources of learning on policy actor policy learning as belief change or
reinforcement in Chapter Three. MNLR models were also used to test the effects of
extremeness of beliefs, policy activities, and sources of learning on policy actor political
learning as advocacy strategy change or reinforcement in Chapter Four. The normality,
skewedness, and relevant assumptions regarding the tests were examined, as well as the
potential for any interactions between independent variables.
CONTRIBUTIONS
Primarily, this dissertation aims to contribute to the policy process literature.
Specifically, this study attempts to improve existing models and theories of policy actor


24
learning regarding policy beliefs and advocacy strategies. In this way this study hopes to
be part of a research effort that advances the theories and measures of learning using the
ACF and policy process theories more broadly (Kraft & Furlong, 2015; Birkland, 2011;
Freeman, 2006). Building on previous ACF applications, this study aims to better
understand factors affecting learning. By examining the role of extremeness of belief,
policy activities, and sources of learning, Chapter Three of this study advances research
trying to understand the factors that shape different forms of social and instrumental
policy learning. In Chapter Four, the effect of policy actors extremeness of belief, policy
activities, and sources of learning on political learning are measured, and this will also
add to the landscape of policy process theorizing. This research will hopefully inform
debates regarding potential revisions to established ACF policy belief and policy learning
hypotheses. Similarly, as ACF scholars attempt to build better models regarding changes
to coalition and individual resources and strategies, the findings here may help improve
our understanding of individual political learning.
Beyond the ACF, this dissertation also hopes to inform other policy process
frameworks. In the sense that subsystem contextual factors shaping policy actor learning
are explored, the findings here may also be useful to scholars using Diffusion of
Innovation models of policy process, in that pathways to policy learning and political
learning are an important causal factor in policy diffusion (Berry & Berry, 2014).
Similarly, scholars using the Institutional Analysis and Development Framework may
find the results here useful in their attempts to explain how different sets of actors learn
differently depending on the context of the policy setting (e.g., rules and norms that may
determine frequency of different policy activities) (Ostrom, Cox & Schlager, 2014).


25
Finally, this dissertation hopes to inform the growing literature examining public policy
through the lens of the Social-Ecological Systems Framework.
Understanding the factors that shape individual policy actor learning is
important in the creation and implementation of environmental policy as it relates to
sustainable management of human-used natural resource systems such as watersheds,
fisheries, and forests (Ostrom, Cox & Schlager, 2014; Weible et al., 2010). These
human-used natural resource systems are embedded in complex, social-ecological
systems (SESs) and contain multiple nested factors across spatial and temporal scales,
which can be compared to the relationships between organisms, organs, tissues, and
individuals cells (Ostrom, 2009). For example, the population living within the
jurisdiction of a city has an environmental footprint including human activities in
buildings, transportation patterns, resource consumption, and waste generation. This
footprint would be determined, at least in some part, by policy actions at the local level
and the aggregated impacts of individual behaviors. But this citys SESs would also be
embedded within the larger context of trans-boundary (cross-jurisdictional)
infrastructure systems, such as electricity generation, as well as broader socio-cultural,
political, and economic conditions (Newell & Cousins, 2014; Ramaswami et al., 2012a,
2012b). Changes occur over time in both the biophysical factors as well as the socio-
cultural factors, and the challenge of our cities and communities being resilient to major
changes, as well as the challenge of sustainable development, is dependent on how we
learn and respond to these changes (Schewenius, McPhearson & Elmqvist, 2014;
Anderies, 2015; Newell & Cousins, 2014; Anderies & Janssen, 2013). This study will
specifically contribute to scholarship and theory building related to policy actor


26
learning and factors that shape learning, and specifically in the context of a subnational
SES where policy actors are managing energy resources and climate policy.
Sustainability has emerged as both a science and a concept that can be used to
organize or reorganize environmental policy discussions. Defined broadly by the United
Nations Brundtland Commission as the ability for a generation to meet its needs without
compromising the needs of future generations, sustainability has become a worthy, if not
essential, goal in our management of complex SESs (Anderies, 2015; Anderies & Janssen,
2013; Dietz et al., 2003). The science and concept of sustainability also provides a lens
through which to view environmental problems, policy alternatives, and learning (Henry,
2009). But striving for sustainability does not mitigate value clashes in the policy process
discussed; it merely provides a new vocabulary and highlights the need for diverse
viewpoints to engage complex environmental problems in adaptive and collaborative
ways. Policy responses to environmental problems such as climate change will require
collective action and cooperation across numerous sectors of the economy and society
more broadly. This will require deliberation among and between various stakeholder
groups that may have much to learn from each other in terms of the social dimensions of
policy problems, such as causal mechanisms, as well as political feasibility of policy
solutions. In other words, the science of sustainability highlights the need for policy actor
policy learning and political learning, and better understanding of learning processes in
the policy process.
Moving toward sustainability demands that scientific and technical data be
examined and weighed along with the values we hold as a society regarding
environmental, economic, and social issues, and that we learn to adapt environment


27
management techniques in ways that can respond to changes in all of these factors
(Anderies & Janssen, 2013; Anderies, 2015; Weible et al., 2010; Dietz et al., 2003). In
this way, learning in the policymaking process is a potential pathway toward collective
action, which is in turn a potential pathway toward sustainable management of common
resources. Thus, if we are to work toward sustainable and adaptive management of SESs
and respond to changes in biophysical and socio-cultural factors over time, we must
understand the influences that shape learning processes (Anderies & Janssen, 2013;
Anderies, 2015; Gerlak & Heikkila, 2011). Even more broadly, by refining models of
individual policy learning and political learning, this dissertation ambitiously hopes to
further the attempts to better understand learning as a potential indicator of societal
benefit of policy processes (Weible, 2014a).
Attempts will be immediately undertaken to distribute via publication the findings
of this dissertation. Chapters Three and Four will be revised as part of the goal to create
two separate articles for potential publication in venues such as the Policy Studies
Journal, Review of Policy Research, or Policy Sciences. Chapter Two may be a candidate
for publication in an environmental, or climate change-oriented practitioner journal.


28
CHAPTER II
SUMMARY ANALYSIS OF CLIMATE AND ENERGY
POLICY ACTORS IN COLORADO
CHAPTER SUMMARY
The creation of effective and sustainable climate and energy policy at any level
requires that policy actors have the ability to learn while adapting to a constantly
changing environment. The objective of this chapter is twofold: first, to understand and
describe attributes of the policy actors involved in climate and energy policy debates in
Colorado, including their beliefs across a number of relevant policy questions, and
second, to determine the extent of policy learning, as defined as a range between belief
change and belief reinforcement, and the extent of political learning, as defined by a
change or reinforcement of advocacy strategies, in the sampled population of policy
actors. Colorado is a good case study because, like more than 30 states in the U.S.,
Colorado launched an initiative to address climate change, which resulted in the
creation of the Colorado Climate Action Plan (CAP) in November 2007. The Colorado
CAP is typical in scale, scope, and goals compared to other states. In the spring of 2011,
a survey was emailed to 793 policy actors involved in climate and energy issues in
Colorado, and 260 individuals returned fully completed surveys. The results of the
survey offer a descriptive portrayal of the diversity in respondents attributes and range
of beliefs regarding the cause, severity, and policy solutions needed to address climate
change. Overall, the results indicated that these policy actors work in a relatively
insular manner in terms of information sources and policy activities. In terms of policy
learning and political learning, policy belief reinforcement and advocacy strategy


29
reinforcement were reported with much higher frequency than policy belief change or
change in advocacy strategy.
INTRODUCTION
The creation of effective and sustainable climate and energy policy at any level
requires that policy actors have the ability to learn over time from experience in the
policymaking process, and be flexible and responsive to changes in the nature of the
policy problem and politics at play. Energy and climate policy actors at all levels of
government make decisions based on available scientific evidence, for instance regarding
the threats posed by climate change. Policy actors ability to generate and use relevant
information may vary, however. Likewise, policy actors are presumably at least aware of
a variety of opinions regarding policy solutions preferred by others in their work, but the
extent to which policy actors actually engage in policy activities with individuals with
contrary beliefs likely varies as well.
Understanding attributes such as resources, activities, network contacts, and
beliefs of policy actors is essential to understand the policy process (Birkland, 2011;
Jenkins-Smith et al., 2014). The two main objectives guide this chapter to: 1) understand
and describe attributes of the policy actors (individuals in the government, private,
nonprofit, and academic sectors) involved in climate and energy policy debates in
Colorado, including their beliefs across a number of relevant policy questions; and 2) to
determine the extent of policy learning, as defined as belief change and belief
reinforcement, and extent of political learning, as defined by a change or reinforcement
of advocacy strategies, in the sampled population of policy actors. Despite years of effort


30
in studying learning in public policy, better definitions and measurements of learning is a
critical task for scholars of policy processes (Heikkila & Gerlak, 2013).
Learning can be thought of as the successive changes or reinforcements in the
beliefs or behaviors of individuals regarding their perceptions of problems and related
policies resulting from experience in the policy process (Jenkins-Smith et al., 2014;
Birkland, 2011). For effective decision-making, the propensity for learning is linked to
policy capacity along with other attributes of policy actors such as policy activities and
beliefs. Policy capacity refers to the analytical and administrative abilities and skills of
the state and local governments, as well other non-governmental organizations (business
and nonprofits, e.g.) to adequately generate and use information to respond to issues in
the energy and climate sector (Elgin & Weible, 2013). As policy capacity will be
measured as one of the policy actor attributes in this chapter, this analysis builds on
previous analyses of policy actors and policy capacity in Colorado (Elgin et al., 2012;
Weible etal., 2012).
In the absence of comprehensive and robust federal climate action planning, a
patchwork of state and local government polices has been created over the last decade.
Like over 30 other U.S. states, Colorado launched an initiative to address climate change,
which resulted in the creation of the Colorado Climate Action Plan in November 2007
(EPA, 2015). This plan called for a reduction of the state emission of greenhouse gases
by 20 percent below 2005 levels by 2020, and a longer-term goal of 80 percent below
2005 levels by 2050 (Colorado Climate Action Plan: A Strategy to Address Global
Warming, 2007). In terms of scale and scope of policies, as well as the shorter-term
emission reduction targets, the Colorado CAP is typical of state CAPs (EPA, 2015;


31
Ramseur, 2007). Given this, Colorado could be thought of as a typical case in terms of
case selection (Gerring, 2007).
Colorado represents a state in a transition to what former Governor Ritter referred
to as a New Energy Economy (Colorado CAP, 2007). The state has a long history of
traditional fossil fuel energy production, and is now currently experiencing large growth
in energy industries, like hydraulic fracturing and unconventional oil and gas
development, in addition to the less carbon emission intensive renewable energy sectors
such as wind and solar (Lyng, 2010). There is reason to believe Colorado policy actors
may be especially sensitive to the intersections of energy policy and the threats posed by
climate change, which will be discussed below. The creation and subsequent
implementation of this plan involves the participation of policy actors with diverse beliefs.
During the participation in this policy process, the policy actors involved will have the
potential to learn regarding the social construction of the problem and the relative
effectiveness of different policies (policy learning), as well as the viability and
palatability of different policy solutions (political learning).
Learning is argued to be a critical component to sustainably managing complex
social-ecological systems (SESs) (Anderies, 2015; Anderies & Janssen, 2013; Dietz et al.,
2003). In that stabilizing and maintaining our climate through sustainable climate and
energy policy is essentially the management of an SES, understanding both policy
learning and political learning in the relevant policy actors is essential. The focus of this
chapter is assessing diversity across a host of policy actor attributes, and the extent to
which policy learning and political learning are occurring, in this policymaking setting
during the implementation of the Colorado CAP. Given that this chapter aims to measure


32
climate and energy policy actor attributes and potential learning this case (Colorado)
would seem to be a positive case, a place where the intended measured phenomena will
exist (Goertz, 2006).
The objectives of this chapter were met using a survey administered in 2011 to
260 policy actors involved in climate and energy issues in Colorado. These policy actors
were asked to answer questions related to their particular training and education,
organizational affiliation, day-to-day work, policy-related activities, and climate and
energy policy-related beliefs. The respondents were also asked about policy learning, the
extent to which their policy beliefs have either changed or been reinforced, and about
political learning, regarding changes or reinforcement of their advocacy strategies. This
chapter does not establish a relationship among variables for explanatory empirical
testing, but rather provides a discussion and description (relative distribution and
patterns) of measured variables in order to better understand the policy actors in the
sample group, and provides the context of climate and energy issues in Colorado. The
Advocacy Coalition Framework (ACF) is used to guide the discussion regarding which
attributes of policy actors should be measured. This chapter will set the stage for the two
following quantitative analysis chapters focusing on policy learning and political learning,
respectively.
THE ADVOCACY COALITION FRAMEWORK
The study of learning can be approached from a diverse set of policy process
theories, models, and frameworks (Birkland, 2011; Freeman, 2006). Some stress the
mechanics of diffusion of new ideas in policymaking (Berry & Berry, 2014), while others
examine the way problems and policies are socially constructed (given collective


33
meaning), and how those constructions affect policy design and vise versa (Schneider &
Ingram, 1993, 1997). Scholars working with the Institutional Analysis and Development
Framework, building on the work of Elinor Ostrom (2005), are developing theories and
models of how institutional design may shape individual and collective learning in
policymaking (Heikkila & Gerlak, 2013). Lastly, the Advocacy Coalition Framework
(ACF) has become a leading framework to directly examine and explain the role of
science and technical information, interaction among and between policy actors with
similar and different goals, individual beliefs and political behaviors, and learning
(Jenkins-Smith et al., 2014; Birkland, 2011; Freeman, 2006).
This chapter is guided by the ACF, which was developed by Sabatier and Jenkins-
Smith after a long history of research on environmental problems and policymaking
(Sabatier 1988; Jenkins-Smith et al., 2014). The ACF was applied to study the policy
actors in the climate and energy policy subsystem4 of Coloradomeaning those who
were active within Colorado and Colorado cities, for example the City and County of
Denver. In applying the ACF in the analysis of a political issue, the focus was directed
toward the attributes of policy actors, for instance resources, activities, and beliefs, in
addition to the extent of learning.
The reason to focus on individual policy actors comes from the ACF argument
that individual policy actor beliefs are the main driver of political behavior and coalesce
like-minded actors together into advocacy coalitions (Jenkins-Smith et al., 2014). These
advocacy coalitions can be groups of legislators, agency officials, interest group leaders,
4 A policy subsystem is the larger context of allied and competing groups (interest
groups, institutions, and governments) and policy actors involved in the policymaking
process in a specific or specialized topic (Jenkins-Smith et al. 2014).


34
and researchers with similar policy beliefs, which are normative and empirical beliefs
bound by the territorial and topical components of the specific policy issue at hand. For
instance, a group of policy actors may share the belief that climate change will have
devastating impacts to communities if left unchecked by government intervention. These
groups or the coalition, to varying extents, may share resources and engage in some
coordination. The findings summarized in this chapter are not designed to assess the
existence of coalitions, but to use the ACF to help assess the diversity of attributes,
including, but not limited to, policy beliefs across the sampled population of policy actors
and to examine the extent of learning that has occurred among these policy actors.
Besides policy beliefs, the ACF postulates that other policy actor attributes are
important to measure alongside the extent of learning. A policy actor attribute is a quality
or characteristic that might vary within the population. For instance the organization
affiliation of policy actors, the level of their respective educations, the amount of time
spent in the policymaking field, and their respective information sources used in their
policy work may be important in the learning process (Sabatier & Jenkins-Smith, 1999;
Jenkins-Smith et al., 2014). Likewise, the extent to which policy actors engage in specific
policy activities, for example coalition building, coordination and collaboration within
and across coalitions, and participation in multi-stakeholder and/or consensus-based
processes, are critical to assess in terms of understanding this population and the extent of
learning that has occurred (Sabatier & Jenkins-Smith, 1999; Jenkins-Smith et al., 2014).
This chapter summarizes policy actors engagement in specific policy activities together
with other policy actor attributes measured in the survey, and the extent of policy and
political learning.


35
POLICY ACTOR ATTRIBUTES MEASURED
This section will detail the concepts and measures from the survey and provide a
short justification of their potential importance in relation to learning, but this chapter
will not be testing any specific hypotheses. Examining the diversity of policy actors
attributes in terms of: demographics, policy activities, policy capacity, information
sources, and policy beliefs will help map the political landscape in this policy subsystem.
The operational measures for each variable (the specific survey questions used to
measure each attribute) are provided in a table in the methodology section below.
Demographics of Policy Actors
Demographic attributes (qualities or characteristics), for example gender,
race/ethnicity, and political orientation were measured to describe the sampled population
of policy actors. Other policy actor demographic attributes may have a more intuitive link
to learning than gender or race/ethnicity, for instance, organizational affiliation (Jenkins-
Smith et al., 2014). Organization affiliation is the type of organization, or sector, in which
the policy actors is predominantly employed. Four major organizational sectors are
represented in the sample of policy actors: government employees, individuals from the
private sector, individuals from the nonprofit sector, and researchers/academics. There is
reason to believe that policy actors organization or company affiliation may be a factor
in shaping learning. For instance, policy actors from government and administrative
agencies may possess more moderate and flexible (or, conversely, less extreme) policy
beliefs than actors from non-profit environmental groups or private sector oil and gas
companies (Jenkins-Smith et al., 2014).


36
Other demographic attributes examined and described in this chapter are the level
of formal education and time (as measured in years) that policy actors have been involved
in climate and energy policy. These attributes were measured on the survey to help
provide an overview of the diversity of whom these policy actors are, to help understand
the politics on this issue. Causal relationships between some of these attributes and
learning will be examined in following chapters.
Policy Activities
Policy activities are defined as any actions or behaviors individual actors engage
in as part of their policymaking work5. The survey asked individuals whether, and how
frequently, they participated in a host of policy activities. Research in policymaking
processes, specifically in the ACF literature, suggests a few potential relationships
between policy actor activities and policy learning (Jenkins-Smith et al., 2014). For
instance, individuals that report they seek advice from those with similar beliefs with
greater frequency than those that seek advice from those with dissimilar beliefs may
suggest the individual has a relatively insular political network, potentially limiting the
new ideas and information that might stimulate belief change (Sabatier & Jenkins-Smith,
1999). Relatedly, individuals that have participated in more facilitated and consensus-
based processes may have been exposed to a variety of viewpoints on energy and climate
policy issues in a professional setting with established rules and norms. These individuals
might then report more belief or political behavioral change, as opposed to reinforcement,
as their political networks are less insular (Jenkins-Smith et al., 2014).
5 Involvement in the decisions of a government or other authority regarding, for example,
laws, regulations, programs, and executive functions (Birkland, 2011).


37
Policy Capacity
Policy capacity has been described as the individual and/or organizational ability
to acquire and use knowledge in policymaking (Howlett, 2009). Therefore, for purposes
of this chapter, policy capacity is defined as the analytical and administrative abilities
and skills of the policy actors involved in a policy process to adequately generate and
use information to respond to issues in the energy and climate sector (Elgin & Weible,
2013). Important factors to consider in measuring policy capacity would be: the
perceived priority of the issue with the respective organization (reflective of institutional
resources and capacity provided by leadership), the extent to which the government had
the knowledge and skillsets needed to respond to policy issues in terms of staff and
resources, and the ability in engage in long-term planning needed in the maintenance of
public policy (Elgin & Weible, 2013). Individuals were asked to report various measures
of their, and their organizations, policy capacity. Policy capacity is examined here as
part of the energy and climate policy landscape, and descriptive of the policy actors
therein, but the remainder of the broader study will focus on examining learning, as
opposed to policy capacity.
Information Sources
Policymaking work requires the acquisition and use of various sources of
information. Information sources could be the different authorities, types of documents,
and materials used by policy actors in their climate and energy policy-related work.
Policy actors information sources are expected to vary in terms of diversity of
information sources and frequency of use of different information sources. Understanding
the frequency policy actors utilize different information sources in their work will help


38
understand the role of scientific and technical information in this policy subsystem.
Policy actors diversity of information sources, or frequency of use of different
information sources may be a factor that impacts learning.
Policy Beliefs
The ACF postulates that individual policy actors seek to influence the policy
process and that the respective individual policy beliefs are the main drivers of their
behavior (Sabatier & Jenkins-Smith, 1999). Therefore, beliefs are important to
understand in an examination of a policy area in terms of some of the politics that may be
at play. Policy beliefs can be defined as a collection of normative and empirical beliefs
spanning the substantive policy topic (in this case climate and energy policy) in a
geographic area (in this case Colorado) (Sabatier & Jenkins-Smith, 1993, 1999).
Policy actors in the climate and energy policy subsystem are expected to have a
variety of beliefs regarding a host of policy related issues. These policy beliefs are
considered to be resistant, but not impossible, to change (Sabatier & Jenkins-Smith,
1999). Understanding the shape and scope of diversity of the policy beliefs in the sample
of the policy actors is essential, therefore, in terms of being able to describe this group,
and to understand what learning has occurred. The survey participants were asked about
their relative agreement or disagreement with six policy belief questions regarding major
aspects of climate and energy policy and policy preferences. Extreme policy beliefs (e.g.,
extreme agreement or disagreement) may vary by organizational affiliation, but also may
be a potential attribute shaping learning. The relationships between organizational
affiliation and extreme beliefs, and the relationship between extreme beliefs and learning,
will both be examined in following chapters.


39
Policy and Political Learning
While no singular conceptual or operational definition of learning across the
social sciences exists, learning is often defined as a normative goal and perhaps a
prerequisite for collaborative and effective policymaking (Henry, 2009; Freeman, 2006).
It is therefore important to think through some different conceptualizations of learning.
Bennett and Howletf s (1992) effort to synthesize theories of learning in public affairs
parsed three major dimensions of learning: 1) who is learning, 2) what is being learned,
and 3) what are the results of learning. For example, policy actors may learn about the
causal mechanisms of climate change, potentially leading to policy change. It is
important also to differentiate between various definitions of what the outcomes or results
of learning might be. For instance, the outcomes of learning may be conceptualized as
broad policy change (Busenberg, 2001), while others suggest the results of learning might
be individual and/or group belief changes that may or may not lead to policy change
(Sabatier & Jenkins-Smith, 1999; Jenkins-Smith et al., 2014). Still others have
differentiated between cognitive and behavioral changes as measures of learning (Gerlak
& Heikkila, 2011). This final differentiation leads to another often-cited typology of
learning in public policy that can be used to explore different dimensions of learning.
Mays (1992) article linking policy learning and failure differentiated two major
aspects of learning in public policy. First, policy learning is concerned with two major
sub-dimensions of learning: 1) social policy learning regarding changes in the social
construction of a policy or problem (such as humans as a driving cause of climate change,
or the relative need for the government to intervene to address climate change); and 2)
instrumental policy learning regarding changes to the viability or effectiveness of


40
different policy implementation designs to reach existing goals (such as the need for a
cap and trade system or a carbon tax to address climate change). The second major
dimension of learning May (1992) parses is political learning regarding changes in
advocacy strategy for specific policy goals (such as a shift in arguments or tactics). This
chapter, as well as the larger study therein, will examine both policy and political
learning.
Learning has been explicitly argued to be especially relevant in the policymaking
process in the ACF theorizing, where it is considered to be an important pathway to
policy change (Sabatier, 1988). In the context of the policy process, the ACF has defined
learning as ongoing adjustments to thought or behavior related to attainment of policy
goals from new experiences and/or evidence (Sabatier & Jenkins-Smith, 1999, p. 123).
From this definition, learning is typically considered to be associated with changes in
beliefs or behavior, but might policy actors also reinforce their policy beliefs or advocacy
behaviors? In other words, individual belief or behavioral change is not the only measure
of learning.
Belief or behavioral reinforcement may occur when the experiences and activities
of policy actors serve to galvanize and strengthen their policy beliefs and political
activities. If inherent value and belief clashes in policymaking lead to intractability, a
reinforcement of beliefs and advocacy tactics would seem to only worsen the sclerotic
policy process. Conversely, perhaps more change in beliefs and advocacy behaviors
during the policymaking process, especially change in a more moderate direction, might
lead to more consensus and generate meaningful policy outcomes. If belief and
behavioral change increased understanding among stakeholders with differing values, and


41
did not lead to more polarization, perhaps sustainable and adaptive climate policy may be
more likely to move forward. While consensus is not the only pathway to policy change,
nor is consensus a normative goal, a better understanding the factors that shape policy
actor learning is needed.
Therefore, delineating between belief change and belief reinforcement, as well as
between change and reinforcement of advocacy strategy seems essential in understanding
the full picture of what kind of policy learning is taking place in this case. While both
belief and behavioral change and reinforcement may be included in a concept of learning
in the ACF, this chapter (as well as the broader study) argues that change and
reinforcement are different kinds of learning, different dimensions of the concept of
learning (Goertz, 2006). Specifically examining policy belief and advocacy strategy
reinforcement as a dimension of learning, in local and state policy actors in the climate
and energy policy field, is one of the significant contributions of the findings described in
this study.
Policy learning is defined here as both belief change and belief reinforcement,
similarly political learning is defined as advocacy strategy change and reinforcement. In
other words, belief and behavioral change, and belief and behavioral reinforcement, are
considered different categories of learning in this study, calling for measuring the relative
occurrence of each, as well as the absence of either belief or behavioral change and belief
or behavioral reinforcement. Reinforcement can be conceptualized as unchanged, but
more firmly held belief or advocacy strategy, further buttressed and supported by
information or experience. This is different from a change in belief, for instance, toward a
more moderate stance, or a change in advocacy strategy, such as a shift in media strategy.


42
In addition to examining the occurrence learning, to better understand and
describe the individuals included in this sample, this chapter will include a description of
measurement of a range of policy actor attributes, along with a discussion of measured
variables to provide justification in later analysis. However, this chapter will not include
an examination of relationships for empirical testing.
CLIMATE AND ENERGY POLICY IN COLORADO
Colorados history of climate and energy policymaking did not begin with the
2007 Climate Action Plan. In 1994 dedicated employees of The Air Pollution Control
Division of the Colorado Department of Public Health and Environment (DPHE) applied
and received over $100,000 in Environmental Protection Agency funding to conduct a
state greenhouse gas emission inventory (Rabe, 2004). This was done so with an attempt
to frame the relatively new policy issue of climate change as linked with long-term air
pollution control and abatement efforts ongoing in the state (Rabe, 2004). Even before
this, the City and County of Denver had demonstrated national leadership by becoming
an initial member of the International Coalition of Local Environmental Initiatives (now
known as ICLEI) urban greenhouse gas emission reduction program (Bulkeley & Betsill,
2003). A 1991 City ordinance and 1995 Mayoral Proclamation joining the Cities for
Climate Protection program, both during the term of then Mayor Wellington Webb,
earned Denver a Clean Cities designation, only the second U.S. city to do so (Bulkeley
& Betsill, 2003). Mayor Webb, who in his attempts to be viewed as an environmental
leader and could be viewed here as a policy entrepreneur, also attempted to frame climate
change issues as part of broader air quality issues in the City (Bulkeley & Betsill, 2003).


43
Politics soon shifted on this issue, however. When an initial state action plan to
reduce greenhouse gas emission was drafted in 1997, the plan called for increased
regulations on the electric utilities sector (Rabe, 2004). This triggered a political backlash
from the coal industry, which at the time dominated the energy generation mix in
Colorado (Rabe, 2004). The result of this backlash was that Colorado became one of 16
states in 1997 and 1998 that passed legislative resolutions expressing: concern over the
local economic impact of greenhouse gas mitigation efforts, criticism of the Kyoto
Protocol (adopted in December of 1997), and opposition to ratification of the
international agreement by the U.S. Senate (Rabe, 2004). While the Colorado legislative
resolution attempting to restrict any state actions designed to reduce greenhouse gas
emission was never turned into binding legislation, it slowed DPHE efforts to enact
modest greenhouse gas emission reduction plans and demonstrated that the energy
industry and other interest groups were capable of framing climate change as an
economic issue (Rabe, 2004).
Now, Colorado represents a state in a transition to what former Governor Ritter
referred to as a New Energy Economy (Colorado Climate Action Plan: A Strategy to
Address Global Warming, 2007). The state has a history of traditional oil production, and
is now currently experiencing large growth in multiple energy industries, like natural gas
development, and less carbon emission intensive renewable energy sectors, wind and
solar energy for instance. Many of these recent changes in the energy portfolio of
Colorado have been due to state and local policy drivers including the Colorado Climate
Action Plan (2007), the Clean Air-Cool Jobs (2010), and the 2004 establishment of the
states renewable energy standard, which was strengthened by Governor Ritter in 2010,


44
opening a new policy landscape and creating new spaces and opportunities for private
sector investment and growth (Lyng, 2010).
There is reason to believe Colorado policy actors may be especially sensitive to
the intersections of energy policy and the threats posed by climate change such as: shorter
and warmer winters creating a thinner snowpack, which may jeopardize the important
skiing-tourism industry, earlier melting of the snowpack with increased spring runoff
leading to significant flash flooding, increased periods of drought and increases in the
number of wildfires, and substantial losses of alpine forests due to pine beetle infestations
(Colorado Climate Action Plan, 2007). Similar to more than 30 states and 1,000 cities in
the U.S., Colorado launched an initiative to address climate change, which resulted in the
creation of the aforementioned Colorado Climate Action Plan in November 2007. This
plan called for a reduction of the state emission of greenhouse gases by 20 percent by
2020. Like other CAPs in the U.S., Colorados climate action plan outlines the package
of policies that will be employed to address climate change by making specific policy
actions toward reducing the GHG emissions. This Colorado Climate Action Plan (CAP)
is similar in scale and scope to the over 30 other state CAPs in the U.S. (EPA, 2015;
Ramseur, 2007).
METHODOLOGY
Research Design
The findings presented in this chapter are based on a survey conducted in the
spring of 2011. The survey was emailed to 793 individuals who were knowledgeable of
and actively involved in climate-related issues and energy policies in Colorado. Two
hundred and sixty (260) respondents filled out the survey completely, for a response rate


45
of 33 percent, and the population reflects a broad and diverse set of policy participants.
The individuals were selected for participation using a combination of a modified
snowball-sampling technique, which began with current and former staff from the
Governors office to see who was involved in the creation of the state CAP. Internet
research, focusing on related policy documents, was also conducted to select additional
individuals from nongovernment organizations and businesses who were involved in
climate and energy policy issues. The fourth tactic used to snowball the sample was to
include in the invitation to be surveyed a request for additional names of stakeholders
involved in climate and energy issues.
Operational Measures of Policy Actor Attributes
Table 2.1 below provides each of the specific survey questions used to measure
each of the policy actor attributes examined in this chapter.
Table 2.1. Operational Measures of Attributes
Operational Measures / Survey Questions: Demographics
Respondents were asked to identify their:
Gender (male or female)
Race/ethnicity (American Indian or Alaska Native, Asian, Black or African American, Hispanic
or Latino, Native Hawaiian or Other Pacific Islander, and White not of Hispanic origin).
How liberal or conservative they consider themselves to be on fiscal policy (on a scale of: very
liberal, liberal, moderate, conservative, and very conservative)
How liberal or conservative they consider themselves to be on social policy (on a scale of: very
liberal, liberal, moderate, conservative, and very conservative)
What best describes their organization (academic/research, business/private sector, government,
media, and non-profit)
Highest level of education attained (not a high school graduate, high school graduate, some
college, bachelors degree, masters or professional degree, and Ph.D./M.D./J.D.)
Years they have been involved in climate-related issues and/or energy policy (on a scale of: <1,
1-5, 6-9, 10-14, 15-20, and >20)
Operational Measures / Survey Questions: Policy Activities
Respondents were asked to report how often they used the following tools and techniques as part of their
work in the past year (on a scale of: daily, weekly, monthly, yearly, and never):
Collaboration with those who share their views on energy & climate goals
Collaboration with those who share their views on energy & climate goals
Facilitation/consensus building (e.g., focus groups, roundtables)


46
Table 2.1: Operational Measures of Attributes (cont.)
Respondents were asked (yes or no) in the past year, which of the following activities did they
participate in:
Participated in coalition building (e.g., networking, information sharing)
Negotiated in a multi-stakeholder consensus-based process
Operational Measures / Survey Questions: Policy Capacity
Respondents were asked to report how often they used the following tools and techniques as part of their
work in the past year (on a scale of: daily, weekly, monthly, yearly, and never):
Community-level impact analysis (e.g., neighborhood surveys)
Political feasibility analysis (e.g., SWOT analysis, polling data)
Risk assessment
Modeling (e.g., climate change scenarios, energy futures analysis)
Environmental impact analysis
Economic and financial analysis (e.g., cost-benefit and economic-impact analysis)
Informal tools and techniques (e.g., brainstorming, problem mapping)
Respondents were asked to respond (on a scale of: very low capacity, low capacity, medium capacity,
high capacity, and very high capacity) to the following question:
Compared to similar organizations, does your organization have adequate knowledge, skills,
and people to respond to climate-related issues and energy policies?
Respondents were asked to respond (on a scale of: much lower, lower, about the same, higher, and much
higher) to the following question:
Compared with other issues that your organization responds to, how much of a priority are
climate-related issues and energy policies?
Respondents were asked (yes or no) in which of the following areas have they received formal training:
Applied research
Modeling
Policy analysis
Policy evaluation
Statistical methods
Trends analysis and/or forecasting
Respondents were asked to indicate their level of agreement (on a scale of: strongly disagree, disagree,
neither agree nor disagree, agree, and strongly agree) with the following statement:
I have timely access to academic literature, peer-reviewed publications and professional
research relevant to climate-related issues and energy policy work.
Respondents were asked (yes or no) in the past year, which of the following activities did they
participate in:
Appraised policy options
Conducted research on climate-related issues and/or energy policy
Consulted with the public
Evaluated policy processes, results, and outcomes
Implemented or delivered policies or programs on climate-related issues and/or energy policy
Informed elected and appointed officials


47
Table 2.1: Operational Measures of Attributes (cont.)
Respondents were asked, with respect to climate and energy issues, to indicate their relative agreement
(on a scale of: strongly agree, agree, neutral, disagree, and strongly disagree) with the following
statements:
I regularly engage in tasks which demand immediate attention (e.g., firefighting)
I regularly engage in tasks which relate to long-term planning (e.g., more than a year)
Urgent day-to-day issues seem to take precedence over thinking long-term
lam provided enough time and resources to undertake tasks and planning that are engaging
for more than a year
lam increasingly consulting with the public as I do my policy-related work
Policy decisions seem to increasingly be those that are most politically acceptable
There seems to be less governmental capacity to analyze policy options than there used to be
My policy related-related work increasingly involves networks of people across regions, or
levels of government
Policy problems increasingly require strong technical expertise
Much of the existing skills and knowledge about climate and energy issues lies outside the
formal structure of government
Those who have more authority in decision-making usually have less specialized technical
expertise
Operational Measures / Survey Questions: Information Sources
Respondents were asked to indicate how often they used the following information sources in their
policy work (on a scale of: daily, weekly, monthly, yearly, and never):
Reports produced/created by their organization
Advice from individuals they agree with
Advice from individuals they disagree with
Reports from non-profits
Personal experience
Budgets and cost data
Academic research
Newspapers and magazines
Reports of other government agencies
Reports from consultants
Industry reports
Online social networks (Facebook, Twitter)
Operational Measures / Survey Questions: Policy Beliefs
Respondents were asked to indicate their relative agreement (on a scale of strongly agree, somewhat
agree, neither agree nor disagree, somewhat disagree, and strongly disagree) with the following
statements:
The severity of predicted impacts on society from climate change are vastly overstated
Human behavior is the principal cause of climate change
Decisions about energy and its effect on climate are best left to the economic market, and not to
the government
An energy and/or carbon tax is required to combat climate change
A cap and trade system of permits for the emission of greenhouse gasses is required to combat
climate change
Government policies to promote renewable energy generation are required to combat climate
change


48
Table 2.1: Operational Measures of Attributes (cont.)
Operational Measures / Survey Questions: Policy Learning
For each of the six policy belief questions above, respondents were asked to what extent their views had
changed or been reinforced:
Only been reinforced
Mostly reinforced with little changes
Balance of reinforcement and changes
Mostly changed with little reinforcement
Only been changed
My views have neither changed nor been reinforced
Operational Measures / Survey Questions: Political Learning
Respondents were asked to what extent their strategies had been changed or reinforced regarding the
way they advocate for climate-related issues and/or energy policy:
Only been reinforced
Mostly reinforced with little changes
Balance of reinforcement and changes
Mostly changed with little reinforcement
Only been changed
Neither changed nor been reinforced
RESULTS
Objective 1: Policy Actor Attributes
The results presented below are broken down by policy actor attribute, policy
activities, aspects of policy capacity, policy beliefs, and advocacy coalitions.
Demographics
The majority of respondents in the sample (54 percent) were female. In terms of
race, the sample was overwhelmingly white (not of Hispanic origin), with 90 percent of
the sample self-identifying in that way. Nearly 50 percent of respondents described
themselves as moderate on fiscal policy, with a relatively even distribution indicating
they were more liberal or more conservative on fiscal policy. Over 70 percent of
respondents reported themselves to be liberal to very liberal on social policy.


49
Survey Respondents by Political
Orientation
Fiscal Policy Social Policy
140 -- ---------------- ----------------
£ 120 1-

s
Very Liberal Liberal Moderate Conservative Very
Conservative
Figure 2.1. Survey Respondents by Political Orientation
Organizational Affiliation
Respondents were asked to identify their organization with the following
question: Which of the following best describes your organization? As seen in Figure
2.2 below: a third of the respondents were from the business/private sector, nearly a third
were government employees, and the other third worked for non-profit organizations
and/or identified as working in the fields of academic/research. Two respondents
(<1 percent) were representatives of the media.


50
87
33%
Organization Affiliation of Survey
Respondents:
260 Policy Actors
2
21%
Figure 2.2. Organizational Affiliation of Survey Respondents
Government officials came from city, state, and federal-level agencies. Researchers
came from colleges and universities, private consulting firms, and government research
organizations. Non-profit organizations consist largely of environmental groups active in
energy and climate-related issues in Colorado. Given the low response rate from media
representatives, they were not included in the analysis below.
Formal Education
In terms of formal education, 53 of 87 (61 percent) of the respondents from the
business sector reported having obtained at least a Masters Professional Degree, post-
secondary professional degrees, or higher. Sixty-five of 79 (82 percent) and 44 of 54
(81 percent) reported the same in the government and non-profit sectors, respectively.


Individuals who identified as researchers/academics had the highest percentage, having
earned a Ph.D., M.D., or J.D. at 51 percent.
51
Respondents by Formal Education
100%
90%
9 80%
| 70%
*3 60%
~ 50%
2 40%
§ 30%
S 20%
^ 10%
0%
Ph.D., MD or JD
Master's or Professional
Degree
Bachelor's Degree
Some College
High School Graduate
Figure 2.3. Respondents by Formal Education
Years Involved
Across all sectors, more than half of survey respondents indicated they had been
involved in climate and energy issues for fewer than 10 years, with the most common
response 1-5 years. In comparison to the other sectors, individuals working for
government agencies reported a shorter time of involvement in climate and energy issues,
with over 50 percent indicating that they had been involved five years or less.


52
Respondents by Years Participating in
Climate and Energy Issues
re
s
a
>
a
e
a>
u
s-
0)
a.
100%
80%
60%
40%
20%
0%
B> 20 Years
15-20 Years
D10-14 Years
6-9 Years
1-5 Years
< 1 Year
Figure 2.4. Respondents by Years Participating in Climate and Energy Issues
Policy Activities
As can be seen in Figure 2.5 below, respondents reported they more frequently
collaborated with individuals with similar views on climate and energy goals than they
collaborated with individuals with dissimilar views on these issues.


53
Frequency of Enagement in Policy
Activities
Collaborate with Those Who Share My Views
Collaborate with Those Who Do Not Share My Views
Participate in Facilitation/Consensus building
90
Daily Weekly Monthly Yearly Never
Frequency
Figure 2.5. Respondents by Frequency of Engagement in Policy Activities
The vast majority of respondents reported collaboration with those who share my
views on climate and energy goals on a much more frequent basis (the most common
answers being daily or weekly) than they reported collaboration with those those who do
not share my views ... (the most common answer being monthly). Participating in
facilitation/consensus building was overall less common. A total of 159 of 260
respondents (61 percent) reported they engaged in facilitation/consensus building (focus
groups and roundtables, e.g.) only monthly or yearly, and 49 respondents (19 percent)
reported they never engage in this policy activity.


54
The results above comport with additional policy activities-related results
regarding two yes/no policy activity questions. First, 63 percent of policy actors answered,
Yes they had participated in coalition building (networking and information sharing,
e.g.) in the past year. Second, just over half (56 percent) indicated Yes, they had
negotiated in multi-stakeholder consensus-based processes. These results together
would seem to indicate that policy actors are more likely to engage in policy activities
within coalitions of individuals with similar views, as opposed to between coalitions of
actors with dissimilar views.
Policy Capacity
The analysis of questions measuring policy capacity indicated that the majority of
individuals reported similar levels of experience and formal training, in a variety of
analytical techniques and use of tools and techniques as part of their work in the previous
year. It is worth noting that when asked directly about policy capacity, government
agency employees reported the lowest levels of organizational policy capacity.


55
"Compared to similar organizations, does
your organization have adequate knowledge,
skills, and people to respond to climate-
related issues and energy policies?"
VI
"re
s
a
>
3
e
e
a>
u

m
a.
100%
80%
60%
40%
20%
0%
Very High
High
Medium
Low
Very Low
Figure 2.6. Extent of Policy Capacity in Respondents Organization
Overall though, policy actors who responded to our survey reported that their
organizations had high policy capacity and were able to address climate change and
energy issues based on their resources. The relative priority of climate and energy-
related issues within an organization may indicate the amount of human and technical
resources (capacity) allocated by leadership to address those issues. Across sectors,
respondents also indicated that climate and energy issues were a higher or much higher
organizational priority compared to other issueswhich is perhaps not surprising given
that participation in these policy issues was the reason these individuals were selected
for the survey.


56
Priority of Climate and Energy-Related Issues
100%
2 90%
= 80%
> 70%
1 60%
- 50%
2 40%
g 30%
£ 20%
a, 10%
0%
Compared to Other Issues in Organization
Much Higher
Higher
About the Same
Lower
Much Lower
Figure 2.7. Priority of Climate and Energy-Related Issues
In terms of the amount of research conducted and accessed, academic and
professional researchers reported the highest abilities to conduct research in terms of
formal training, while government employees reported the second highest ratios of formal
training.
Table 2.2. Percentage of Respondents That Received Training in Applied Research
Percentage of Respondents That Received Training in Applied Research
Research / Academic 66%
Business 30%
Government 37%
Non-profit 18%
Individuals from all sectors reported having timely access to academic and
professional research on climate and energy policy, but government employees reported
so less frequently than the academic/research of the business sectors.


57
"I Have Timely Access to Professional
Research" by Sector
Strongly Agree
Agree
Neither Agree nor Disagree
Disagree
Strongly Disagree
Figure 2.8. Respondents Reported Access to Professional Research
Regarding the ability to measure the opinions and preferences of the public,
interest groups, and major policy players (an important aspect of policy capacity
discussed above)compared to the other sectors, government agency employees
reported the highest frequency of use of community-level impact analyses such as
neighborhood surveyswith the most common response indicating yearly. In terms of
another important component of policy capacity, the ability to communicate to
stakeholders, the survey results indicate the majority of government employees
(51 percent) reported that their policy work increasingly involved consulting with the
public, and 71 percent reported they participated in coalition building (e.g., networking or
information sharing).
Lastly, summarizing the final set of policy capacity-related results of Yes/No
questions measuring the ability to articulate medium-and long-term priorities: a


58
majority of government agency employees (61 percent) indicated they regularly
engaged in long-term tasks and 48 percent also indicated they were not provided with
the necessary resources and time to engage in long-term planning. Additionally, 73
percent of government employees indicated short-term issues took precedence over
long-term thinking.
Information Sources
The survey respondents were asked how frequently they used 12 sources of
information in their climate and energy-related policy work. Table 2.3 below presents a
summary of responses organized by descending values of daily usage, helping to identify
what information sources were used most frequently. Two noteworthy items: more than
50 percent of respondents indicated they used personal experience daily in their policy
work, and more than 50 percent of respondents indicated they never used online social
networks in their policy work. Also standing out: besides online social networks,
respondents indicated they used each source of information at least yearly 90 percent of
the time. In other words, besides online social networks, only 10 percent or less of the
time did respondents on average indicate they never used any individual information
source. The summary table below thus supports the following: a large percentage of
respondents used the vast majority of given information sources at least to some extent.
The summary total does not include a sum column or row due to the fact that sometimes
one or two of the 260 respondents skipped (did not answer) one of the information
sources, but the total for each source ranged 256-260.


59
Table 2.3. Respondent Information Use Summary
Daily Weekly Monthly Yearly Never
Personal experience 57% 18% 16% 5% 4%
Newspapers and news magazines 42% 26% 22% 6% 4%
Advice from people you agree with 31% 34% 29% 4% 2%
Reports created by your own organization 18% 24% 31% 17% 10%
Budget and cost data 18% 26% 31% 15% 10%
Advice from people you disagree with 17% 22% 45% 11% 5%
Academic research 16% 23% 42% 14% 5%
Reports from industry 13% 27% 37% 19% 4%
Online social networks 11% 15% 12% 7% 54%
Reports from other governments 9% 26% 43% 15% 7%
Reports from consultants 8% 22% 41% 20% 9%
Reports from non-profits 6% 26% 40% 19% 9%
Other trends that include advice from people you agree with and reports
created by your own organization were reported by respondents as used more frequently
in their policy work than, for instance, advice from people you disagree with, or reports
from the various outside organizations queried. This, combined with the prime use of
personal experience, would seem to indicate a relatively insular set of information
sources being utilized by the policy actors in this subsystem.
Policy Beliefs
The policy actors were asked to demonstrate their policy beliefs about climate
change and energy policy-related issues by indicating their relative agreement or
disagreement with six policy statements. The results are shown in Figure 2.9 below. The
majority of respondents strongly agreed with the statement about the need for policies to
promote renewable energy, with the chart of policy beliefs below organized by
descending levels of strong agreement with the six statements.


60
To summarize, the majority of participants reported they strongly agree or agree
that: 1) the impacts of climate change are not overstated; 2) human behavior is the
principal cause of climate change; 3) decisions about energy and its effect on climate are
best left to the government, as opposed to the market; and 4) government polices to
promote renewable energy generation are required to combat climate change. Compared
to these four policy beliefs, there was greater diversity of opinions on the need for a
carbon tax or a cap and trade system to combat climate change, but the majority of
respondents strongly or somewhat agreed with the need for the former, and respondents
were more neutral (more reported neither agree nor disagree) on the need for the latter.


61
Policy Beliefs
Strongly Agree Somewhat Agree Q Neither Agree Nor Disagree
Q Somewhat Disagree Strongly Disagree
Renewable Change Combat Climate Combat Climate are Overstated are Best Left to
Energy Change Change the Market, and
Generation are Not to the
Required to Government
Combat Climate
Change
Figure 2.9. Policy Beliefs
Advocacy Coalitions
Elgin and Weible (2013) conducted a different, but related, analysis of this same
sample of Colorado climate and energy policy actors. While extensive mapping and
exploration of the advocacy coalition in this subsystem is not an objective of the analysis
presented in this chapter, some discussion of this is warranted. Analyzing the results of
the six policy belief questions using cluster analysis and silhouette means, Elgin and
Weible (2013) found that the actors fell into two advocacy coalitions based on belief
systems. A larger proclimate coalition of 205/260 (79 percent) policy actors believed


62
that climate change impacts were serious and should be addressed by government
intervention, and were supportive of policies aimed at doing so (Elgin & Weible, 2013).
The anticlimate coalition consisted of 55/260 (21 percent) policy actors with
fundamentally opposing policy beliefs (Elgin & Weible, 2013). Besides vastly different
policy beliefs across coalitions, two other significant differences related to coalition
membership were observed. First, the proclimate coalition had a majority of its members
from the government sector, while the majority of anti climate coalition members were
private sector employees. Second, the proclimate coalition members identified
themselves as significantly more socially and fiscally liberal than the anticlimate
members, who identified themselves as more conservative socially and fiscally (Elgin &
Weible, 2013).
Despite opposing policy beliefs and these differences in attributes among policy
actors across coalitions, the coalitions themselves were found to be similar in many ways
(Elgin & Weible, 2013). The coalitions had similar levels of individual education,
experience, and training. Policy capacity levels in terms of knowledge, skill, resources,
and organizational priority of climate and energy issues were similarly high in both
coalitions. Comparatively, very similar policy activities and advocacy strategies were
used by the two coalition members. Elgin and Weible (2013) concluded that, given the
proclimate coalition was proportionately larger, and that Colorado supports a Climate
Action Plan, there is some reason to believe this coalition was in a stronger position, and
had some policy goal attainment success. That said, the researchers make clear that any
claims regarding relative influence and absolute success of the two advocacy coalitions in
the subsystem would need additional evidence (Elgin & Weible, 2013).


63
Objective 2: Learning
Results presented below provide details regarding policy actors policy learning
and political learning in this subsystem.
Policy Learning
Policy actors were asked to demonstrate their policy learning, defined here as
their responses to a scale between reinforcement of beliefs and change in beliefs, but also
including neither change nor reinforcement of beliefs. As can be seen in Figure 2.10
below, in terms of policy learning, policy actors tended to report belief reinforcement
with much greater frequency than belief change, but neither change nor reinforcement
was more common than belief change.


64
Policy Learning Across the Six Policy
Belief Questions
BThe Predicted Impacts from Climate Change are Overstated
Human Behavior is the Principal Cause of Climate Change
Decisions about Energy & its Effect on Climate are Best Left to the Market...
An Energy and/or Carbon Tax is Required to Combat Climate Change
A Cap & Trade System Is Required to Combat Climate Change
H Government Policies to Promote Renewable Energy Generation are Required...
120 7------------------------------------------------------------------
My Views Have My Views Have My Views Have My Views have My Views Have My Views Have
Only Been Been Mostly Had a Balance of Mostly Changed Changed Only Neither Changed
Reinforced Reinforced With Reinforcement & With Little Nor Been
Little Changes Changes Reinforcement Reinforced
Figure 2.10. Policy Learning Across the Six Policy Belief Questions
Political Learning
Respondents were asked to demonstrate their political learning, defined here as
their responses to a scale between reinforcement of advocacy strategies and change in
advocacy strategies, but neither change nor reinforcement was included. As can be seen
in Table 2.4 below, like with policy learning, in terms of political learning, respondents


65
tended to report reinforcement of advocacy strategy with much greater frequency than
change in advocacy strategy, though neither change nor reinforcement was common.
Table 2.4. Political Learning in Survey Respondents6
Category of Political Learning Regarding Advocacy Strategies Number of Respondents
Only Been Reinforced 59
Mostly Reinforced with Little Changes 59
Balance of Reinforcement and Changes 76
Mostly Changes with Little Reinforcement 10
Only Been Changed 4
Neither Changed nor Been Reinforced 43
CONCLUSIONS
The concept of learning is central to many approaches to studying public policy
and policy processes, and particularly so within the ACF (Freeman, 2006; Birkland,
2011; Heikkila & Gerlak, 2013). Recent meta-studies of learning (see Murro & Jerry,
2008; Reed et al., 2010; Crona & Parker, 2012) agree that despite the lack of a single
definition or measure of learning across the social sciences, learning is often cited as a
goal and perhaps a prerequisite for effective governance and sustainability. It is therefore
important to measure the extent of policy actor learning, and the various forms that
learning may take, when examining policy subsystems and surveying policy actor
attributes in hope of better understanding policy processes.
The first major objective of this chapter was to understand and describe attributes
of the policy actors involved in climate and energy policy debates in Colorado, including
their beliefs across a number of relevant policy questions.
6 Note: the total response count for the question measuring political learning was 251,
slightly below the 260 total for most of the previous questions examined. This may be in
part due to the location of the question toward the end of the survey.


66
The results indicate that this sample of policy actors varies across many attributes
including organizational affiliation and a number of policy activities, as well as across the
policy beliefs examined. Some of the findings support claims made by the ACF regarding
policy actors working within in a policy subsystem, detailed below.
One third of the sampled population were government employees, another third
were from the private sector, and individuals from the non-profit sector and
academic/research field made up the final third. This organizational diversity of policy
actors is consistent with the ACF argument and findings of previous ACF applications
that policy subsystems involve more than just government employees in policymaking
(Jenkins-Smith et al., 2014). The results of the survey show that the policy actors
surveyed were essentially evenly split gender-wise, predominantly white (not of Hispanic
origin), and well educated.
The vast majority of respondents reported more frequent collaboration with policy
actors with similar views than collaboration with policy actors with dissimilar views.
Individuals reported specific participation in facilitated consensus building even less
frequently than collaboration. Engagement in facilitated consensus building (focus groups
and roundtables, e.g.) was predominantly done on only a monthly or yearly basis, and
about a fifth of the respondents reported they never engage in this policy activity at all.
These results together would seem to indicate that policy actors are more likely to engage
in policy activities within coalitions of individuals with similar views, as opposed to
between coalitions. This may support the findings of a previous ACF application that
coalitions of policy actors working together to advance policy goals may be shaped by
the nature of perceived opponents and allies (Henry, Lubell & McCoy, 2011). The


67
frequency of information sources used by the respondents indicates diverse, but also
relatively insular and fragmented policy information utilization.
In terms of policy capacity, results indicate that policy actors working in this
climate and energy policy system believe government has a mixed level of policy capacity
in this area. That said, there was consensus that the policy capacity in the public sector on
these issues needs to be increased. Given the target population and purposive sampling
process, our respondents were very likely knowledgeable individuals on this matter.
In terms of policy beliefs, the results indicate a range of views, but broadly
speaking, the majority of respondents agreed the effects of climate change are serious,
that humans are causing climate change, that the government should be involved in
climate and energy policy and, specifically, in the need for policies promoting renewable
energy generation. Views on the need for a carbon/energy tax or a cap and trade system
were more mixed. This seems consistent with established ACF hypotheses that there is
less consensus on specific policy preferences than on beliefs central to the perception of
the problem; however, a recent review of ACF applications found mixed support for
these hypotheses (Jenkins-Smith et al., 2014).
The second major objective of this study was to determine the extent of policy
learning, as defined as a range between belief change and belief reinforcement, and
extent of political learning, defined here as a range between a change and reinforcement
of advocacy strategy over time in the sampled population ofpolicy actors. In general,
across all six major policy beliefs, reports of policy belief, and advocacy strategy
reinforcement were much more common than reports of policy belief or advocacy
strategy change.


68
The greater incidence of belief and behavioral reinforcement may be of concern,
as it potentially signals that the individuals involved in policy debates in this context are
becoming more intransigent, potentially resulting in their reduced ability to promulgate
effective and adaptive climate change in the near future (Meijerink, 2005; Litfin, 2000).
That said, a compromise of beliefs is not the only pathway for policy change. Relatedly,
the fact that most policy actors reported relatively insular networks of advice and
information sources, as well as relatively less collaboration with individuals with
dissimilar views or participation in facilitated multi-stakeholder consensus-building, may
indicate a paucity of dialogue between policy actors with different beliefs on the relevant
problems and potential policies. This suggests that the subsystem may have trouble
building the policy learning and political learning over time needed to create sustainable
climate and energy policies (Jenkins-Smith et al., 2014). When this is considered along
with the greater prevalence of belief reinforcement as opposed to belief change among
policy actors, this becomes even more troubling. Again, the researcher is not suggesting
consensus is a de facto superior policy process outcome, but normative arguments
regarding a link between learning and sustainable policymaking are made elsewhere
(Henry, 2009). If a greater effort can be made to increase attendance and meaningful
engagement in roundtables and other facilitated consensus building activities, perhaps
these trends can be mitigated or reversed.
It is important to acknowledge the limitations of these data when attempting to
generalize these findings in the Colorado climate and energy policy system to other
policy arenas or geographic areas. For instance, this analysis considered all government
levels (state and city for instance) and agencies together. Relatedly, given that Denver is


69
the capital city (higher population and greater connectivity to state level government
issues), the total number of local government actors is skewed toward the City and
County of Denver employees, and less so on small communities and other cities in
Colorado. Further research is required to unpack the difference between different sub-
populations of government employees, and a similar argument could be made for the
business sector, academic/researchers, and non-profit employees. Moreover, like all self-
reported opinion-based research, the results represent reports potentially limited or biased
by our sampling design or the answers provided.
This chapter presents findings indicating this sample of policy actors displays
diversity across many attributes including organizational affiliation, policy activities, and
among the six policy beliefs measured. Some of these findings bolster assumptions and
arguments made by the ACF and corroborate some previous applications regarding
policy actors, beliefs, and various forms of learning (Jenkins-Smith et al., 2014; Henry,
Lubell & McCoy, 2011; Meijerink, 2005; Litfin, 2000). Additional analysis is required to
explore potential relationships between policy actor attributes and possible effects on
policy actor learning.


70
CHAPTER III
FACTORS SHAPING POLICY LEARNING:
A STUDY OF POLICY ACTORS IN SUBNATIONAL
CLIMATE AND ENERGY ISSUES
CHAPTER SUMMARY
The creation of effective and sustainable climate and energy policy at any level
requires that policy actors have the ability to learn. Information and experiences are
interpreted in ways that may change or reinforce the beliefs of the individuals and groups
engaged in the policy process. This change over time has been defined as policy learning,
and the concept of learning has long played a central role in the theories and frameworks
used to understand policy processes. Findings described in this chapter aim to contribute
to the theoretical and methodological understanding of individual learning in the policy
process by explicitly examining belief change and belief reinforcement as products of
policy learning, measuring both, as well as measuring the absence of either. Analysis
described in this chapter examined several factors associated with policy learning
including: policy actors extremeness of beliefs, the extent to which policy actors engage
in policy activities such as collaboration and advice seeking within and between belief
coalitions, and facilitated consensus-building processes. Similar to many U.S. states,
Colorado is experiencing changes in energy production patterns regarding oil and gas
development and renewable energy generation. Like more than 30 states, these changes
are happening in the context of an initiative to address climate change, the Colorado
Climate Action Plan launched in November 2007. The objective of this chapter is to use
the lens of the Advocacy Coalition Framework (ACF) to help examine some of the factors


71
that promote and shape policy learning in policy actors involved in climate and energy
policy debates in Colorado regarding their beliefs across a number of relevant policy
questions. In the spring of 2011, a survey administered to climate and energy policy actors
in Colorado measured policy learning and the factors that may shape policy learning in
these policy actors. The results indicate that extreme beliefs are associated with belief
reinforcement, relative to policy actors with more moderate beliefs, and that collaboration
with individuals with differing policy views is associated with belief change.
INTRODUCTION
In democratic societies, the policymaking process can involve the clash of
competing values leading to prolonged political intransigence. This in true across many
policy areas, but can be especially true in environmental policy processes where policy
alternatives are explicitly value-laden. In these cases a diversity of policy beliefs held by
individuals may exist across a variety of factors such as: the role of ecosystem protection,
economic development, and social-equity issues. Acquired information and experience
are interpreted in ways that may change or reinforce the beliefs of the individual actors
engaged in the policy process. This change or reinforcement over time has been defined
as a learning, and the concept of learning has long played a central role in the theories
and frameworks used to understand policy processes (Heclo, 1974; Bennett & Howlett,
1992; May, 1992; Sabatier & Jenkins-Smith, 1993; Gerlak& Heikkila, 2011).
In The Science of Muddling Through, Charles Lindblom (1959), an important
early contributor to the study of public policy, offered the idea of policymaking as a
process of incrementalism, a sequence of successive limited comparisons that allow
decision makers to engage in learning from previous policy design attempts and failures.


72
These successive comparisons occur in the policy process, defined by Sabatier (2007,
p. 3) as the process by which .. problems are conceptualized and brought to the
government for solution; governmental institutions formulate alternatives and select
policy solutions; and those solutions get implemented, evaluated, and revised. In
Sabatiers terms, policy actor learning occurs as different policy ideas are brought
forward and compete in the policy process. The study of policy processes can now be
approached from a diverse set of theories and frameworks (Cairney & Heikkila, 2014),
and the concept of learning remains an important component to many of these theories
and frameworks (Heikkila & Gerlak, 2013; Freeman, 2006). But there is still much to
examine about the specific approaches to individual learning in policy processes. For
instance, different factors may shape policy learning in different ways. In other words,
there may be different products (or results) of policy learning, such as policy belief
change or belief reinforcement, based on the attributes and experiences (the learning
process) of individual actors involved in the policy process.
In terms of policy learning as belief change and belief reinforcement, some of the
factors that have been recognized as shaping learning in policy processes are the
uncertainty of relevant scientific and technical information, problem definition, and
policy alternatives (Mazur, 1981). Beliefs may also change, or be reinforced, due to
internal factors, for example limited cognitive abilities, perceptual filters, or preexisting
belief structures which may affect information processing (Simon, 1985). Learning may
also be shaped into different forms or products, such as belief change and belief
reinforcement, from interactions and activities between and among allies and individuals
from groups that advocate different policy goals (Sabatier & Jenkins-Smith, 1993, 1999).


73
Specifically, the extent to which policy actors engage in policy activities such as
collaboration, coordination, and consensus building within and between advocacy
coalitions may also affect information acquisition and potentially shape learning in the
direction of belief change or reinforcement (Sabatier & Jenkins-Smith, 1993, 1999;
Jenkins-Smith et al., 2014).
Better understanding the factors that affect individual policy learning is an
important aspect toward gaining a more complete picture of how policy actors learn in
the policy process. This chapter uses the Advocacy Coalition Framework (ACF) as a lens
to view the processes and products of learning. Factors such as extremeness of beliefs,
different policy activities, and sources of learning may shape learning in different ways
and lead to various forms of learning. Specifically, the role of these variables will be
tested to determine effects on individual policy actor policy learning, defined here as
including belief change, belief reinforcement, and combinations of changes and
reinforcement. This leads to the research question of this chapter:
What factors affect policy actor belief change and reinforcement in a local and
state level energy and climate policy subsystem?
THEORY & HYPOTHESES: POLICY LEARNING AND
THE ADVOCACY COALITION FRAMEWORK
The Advocacy Coalition Framework (ACF), developed by Sabatier and Jenkins-
Smith, has become a leading framework to examine, among other policy processes,
individual policy actor belief and belief change (Birkland, 2011). The ACF builds from a
boundedly rational model of the individual policy actor, possessing perceptual filters
stemming from belief structures, spending time and resources gaining knowledge of


74
specific policy areas, and building political networks in order to affect the development
of public policy (Sabatier, 1988; Jenkins-Smith et al., 2014). Individual policy beliefs7,
defined as a collection of normative and empirical beliefs spanning the substantive policy
topic (in this case climate and energy policy) in a geographic area (in this case Colorado),
the ACF argues, pull actors with similar beliefs together into advocacy coalitions
(Sabatier, 1988; Sabatier & Jenkins-Smith, 1993, 1999). Advocacy coalitions are groups
of different policy actors (e.g., interest group representatives, lawmakers, agency officials,
or researchers), who share policy beliefs and, to at least some extent, collaborate in their
attempts to influence the policy process (Jenkins-Smith et al., 2014). The policy actors as
individuals, as opposed to their advocacy coalitions, will be the unit of analysis for this
chapter, and broader dissertation study, but comparisons will be made between policy
actors with different self-reported belief systems.
A 2009 comprehensive meta-analysis of applications of the ACF found that some
studies have shown policy learning and individual belief change within and across
coalitions to have potential to drive policy development and change over time (Weible et
al., 2009). But previous ACF applications of learning have been inconsistent in the
conceptualization and measurement of this concept and it is considered an aspect of the
ACF deserving of greater practical attention and theoretical innovation (Jenkins-Smith
et al., 2014; Weible, 2011). This chapter attempts to follow the advice of Jenkins-Smith
et al. (2014, p. 205) to improve the ACF and make clear that both belief change and
belief reinforcement are conceptually considered policy learning products, delineate and
7 As this application of the ACF is not examining differences between deep core beliefs,
policy core beliefs, and secondary beliefs, the phrase policy beliefs will be used.


75
measure each clearly, and examine what factors might shape policy learning in either of
these directions.
Theoretical Emphasis
The ACF provides a number of causal relationships and hypotheses regarding
policy beliefs and learning (for a full list of hypotheses see: Jenkins-Smith et al., 2014).
There are three major theoretical emphases of the ACF which are typically the focus of
applications in the literature: 1) advocacy coalitions (mapping the structure, membership,
resources, and changes over time); 2) policy change (examining and explaining the policy
developments over time and the role of coalitions therein); and 3) the concept of policy
learning or policy-oriented learning. The third of these theoretical emphases will be the
focus of this chapter, and broader dissertation study.
To situate this chapter in proper context, a brief discussion of the study of
learning in disciplines other than public affairs is necessary. As Muro and Jeffrey (2008)
point out, various and diverse fields such as social psychology and neuroscience have
been struggling to understand learning for decades. Despite a great deal of studies, it is
difficult to make comparisons across much of the work, in part because of different
assumptions made related to the nature of learning (Muro & Jeffrey, 2008). The concept
of learning is multidimensional, and this can lead to confusion between the concept of
learning itself and its potential outcomes, for instance (Reed et al., 2010). In other words,
is simply acquiring new knowledge enough to constitute learning, or must a deeper
behavioral change result as well to indicate learning has occurred?
Studies that assume new knowledge must be incorporated and utilized by
individuals to constitute learning are making underlying assumptions that other


76
researchers do not (Crona & Parker, 2012). For instance, a different approach to the
question posed above is to ask if individuals in social situations are capable of learned
behavior through conditioning, with no additional information acquisition (Muro &
Jeffrey, 2008). The learning context, and the motivations for learning are also critical
factors to consider when comparing studies. Recent meta-studies of learning (see Muro &
Jeffrey, 2008; Reed et al., 2010; Crona & Parker, 2012) agree that while no singular
definition, conception, or operationalization of learning exists across the social sciences,
learning is often cited as a normative goal and perhaps a prerequisite for adaptive
governance. It is therefore important to delineate the different dimensions of learning and
the factors that might shape policy actor learning.
Heclo (1974) is commonly cited in public policy literature as an early and
important contributor to the study of learning in policymaking, and this work indeed
influenced Sabatiers early theorizing of the ACF relating learning to policy change. But
there is of course diversity in the way learning is conceptualized in the policy process
literature. Some make distinctions between cognitive versus behavioral changes (Gerlak
& Heikkila, 2011), while others look for organizational changes in collective governance
settings (Crona & Parker, 2012). The contextual factors that describe how social
environments and networks can cultivate learning are also important to explore (Reed et
al., 2010), as are broader descriptions of the political environment (Gerlak & Heikkila,
2011). Likewise, the role of learning has evolved into a goal, sometimes explicitly in
collaborative government literature (Muro & Jeffrey, 2008) and the outcomes of learning
may be conceptualized simply as policy change (Busenberg, 2001).


77
The concept of learning is central to many theories, models, and frameworks used in
studying public policy and policy processes (Freeman, 2006; Birkland, 2011; Heikkila &
Gerlak, 2013). Learning, as a part of the policy process, is particularly relevant within the
ACF, as evidenced by the use of the word learning in the title of the seminal article
introducing the ACF (Sabatier, 1988). Within the ACF, learning has been defined as:
relatively enduring alterations of thought or behavioral intentions that result from
experiences and/or new information and that are concerned with the attainment or revision of
policy objectives (Sabatier & Jenkins-Smith, 1999, p. 123). However, across applications of
the ACF, learning has been variously and poorly defined theoretically (Weible et al., 2011).
While dozens of applications of ACF have affirmed that opposing advocacy coalitions are
formed and maintained based on stable policy beliefs among individuals within policy
subsystems (Weible et al., 2011; Weible et al., 2009; Zafonte & Sabatier, 1998), learning has
been the least explored of the three major ACF theoretical emphases (Jenkins-Smith et al.,
2014). One of the reasons for this is that there is little agreement on clear conceptualization
and operationalization of learning. Thus, systematic comparison between different products
of learning, and the various processes of learning that shape different products, has been a
difficult and often ignored or avoided endeavor (Heikkila & Gerlak, 2013).
In the studies conducted over the previous decades, scholars working within the
ACF have found policy learning to be more likely in a context where conflict is at
intermediate levels and focused on specific policy instruments or variations, and when
there are forums with established professional norms (e.g., mediated roundtables) for
individuals in opposing coalitions to collaborate on policy development (Lester &
Hamilton, 1987; Sabatier & Brasher, 1993; Eberg, 1997; Ellison, 1998; Lubell, 2003;


78
Meijerink, 2005; Larsen et al., 2006). While the strength of this past research has been to
better understand policy subsystems and collaboratives where policy change may be
more likely, a limitation is that policy change was often conflated with the concept of
policy learning. Specific models of policy actors learning processes or products have not
been tested as often as the larger policy system attributes, such as the level of conflict.
One of the central challenges in studying learning, at large and when using the
ACF, is a lack of consistency and clarity in how learning is definednot just
conceptually, but also operationally. For example, individual belief change, coalition
change, and policy are variously provided as theoretical indicators of learning in ACF
applications (Weible et al., 2009). In the ACF literature, as well as the policy process
literature more broadly, one or more of these three indicators of learning are often
provided as evidence of policy learning, but rarely is this explicitly linked to a specific
conceptualization of policy learning (Heikkila & Gerlak, 2013; Weible et al., 2011). Even
fewer attempts have been made to examine policy learning (even belief change)
empirically using survey data; instead, researchers have often relied on assessments of
policy change over time from, for example, unsystematic content analysis of existing
documents or historical analysis (Weible et al., 2011, 2009).
This chapter specifically conceptualizes and defines belief reinforcement in
addition to belief change as evidence of learning. Reinforcement is another form of
cognitive change. Belief reinforcement may take the form for instance, of a more firmly
supported policy belief, more buttressed by new information or experiences. Reported
cognitive, as opposed to behavioral, changes signify the results (or products) of policy


79
learning for the purposes of this dissertation, as reported in this chapter8. While both
belief change and belief reinforcement may be included in a concept of policy learning in
the ACF, this chapter includes the argument that they are different kinds of learning. In
other words, belief change and belief reinforcement could be considered oppositely
charged poles within the concept of policy learning, to use an analogy from Goertz (2006).
Policy learning, defined here, includes belief change, belief reinforcement, and
different combinations of these categories, such as policy belief that have been mostly
reinforced, with some changes or a balance of change and reinforcement. Therefore,
policy actors reporting neither belief change, nor belief reinforcement, are indicating not
learning or nonlearning in terms of a lack of cognitive change. In other words, no
change or reinforcement of beliefs could be evidence of nonlearning (the absence of
cognitive change). Recent meta-analysis of learning in public policy have pointed out that
the notion of nonlearning is a conceptually difficult area (Heikkila & Gerlak, 2013;
Gerlak & Heikkila, 2011). In keeping with this studys attempt to be clear on
conceptualization and measurement of learning, for the purposes of this chapter, the lack
of cognitive change or reinforcement will be considered nonlearning.
Largely untested, according to Jenkins-Smith et al. (2014), ACF hypotheses do
exist relating policy actors beliefs to learning. For instance, it is suggested policy actors
may display variation in the extremeness of policy positions they espouse. Further, more
extreme (less moderate) views of policy actors, regarding specific policy issues, may
make policy actors less flexible to change in their beliefs related to those issues. The ACF
8 While most of the learning being examined in this dissertation relates to belief change
or reinforcement (cognitive changes), individuals were asked about changes in advocacy
strategies used to forward policy goalsthis represents a behavioral change and the
investigation into this learning is presented in a later analysis.


80
arguments are based on the largely accepted claim that individual perceptual filters might
bias information assimilation in ways that could lead to confirmation bias and belief
reinforcement (Lord et al., 1979). These infrequently examined ACF hypotheses have
some mixed support in the literature (Jenkins-Smith et al., 2014). As the objective of this
chapter is to further refine ACF hypotheses and models of individual policy learning, this
chapter will present findings from analyses that test the following hypothesis, as adapted
from Jenkins-Smith et al. (2014).
Extreme Belief Hypothesis:
Hypothesis 1: Policy actors with more extreme policy views are more likely to
reinforce their beliefs than change their beliefs.
The ACF argues that policy activities, for example, interactions between and
among individuals that advocate similar policy goals and/or those that advocate different
policy goals, may affect policy learning (Sabatier & Jenkins-Smith, 1993, 1999). For
instance, the extent to which policy actors engage in policy activities such as seeking
advice, collaboration, and consensus building within and between advocacy coalitions
may affect information acquisition. Specifically, if policy actors activities are more
insular in nature in terms of interacting more exclusively, e.g., seeking advice more
frequently from individuals who share their beliefs on climate and energy policy, this
could potentially shape policy learning through mechanisms like confirmation bias (the
process of selection of information that confirms existing preconceptions), in the
direction of belief reinforcement (Sabatier & Jenkins-Smith, 1993, 1999).
Conversely, policy actors engaging in activities that expose them to more diverse
ideas, for instance those that engage in more frequent collaboration with individuals with


81
dissimilar policy beliefs or objectives, may report policy learning shaped in the direction
of belief change. This would be consistent with the finding of Leach et al. (2013), that
new information acquired in collaborative processes also has the potential to change
policy beliefs. Beyond collaboration with those with dissimilar beliefs, the ACF contends
that policy actors participation in facilitated consensus-based process (e.g., mediated
roundtables) and constructive cross-coalition activities may expose them to diverse ideas
on the conceptual problem and possible policy solutions, and potentially shape learning
in the direction of belief change. These arguments lead to three policy activity-related
hypotheses to be tested, also adapted from Jenkins-Smith et al., (2014).
Policy Activities Hypotheses:
Hypothesis 2: Policy actors that seek advice more frequently from those with
similar beliefs are more likely to reinforce their beliefs, rather than change their beliefs.
Hypothesis 3: Policy actors that collaborate more frequently with those with
dissimilar beliefs are more likely to change their beliefs, rather than reinforce their beliefs.
Hypothesis 4: Policy actors that have participated in more frequent facilitated
consensus-based processes are more likely to change their beliefs, rather than reinforce
their beliefs.
CLIMATE CHANGE POLICY
The Intergovernmental Panel on Climate Change, and the broader scientific
community, unequivocally claim and demonstrate that, without intervention, climate
change will have devastating impacts on human communities (Giddons, 2011; IPCC,
2013, 2014; Stern, 2007). Until very recently, the federal government in the U.S. has
done very little promulgating meaningful climate legislation. This vacuum has provided


82
the space for a surge of policy action at the subnational level, where localities and
counties serve as laboratories for the creation, implementation, and examination of
promising programs and legislation. Thus, the study of subnational climate and energy
policy is becoming important among political scientists and policy scholars struggling to
understand the actors and processes in place in these subnational policy subsystems.
Subnational Climate Policy Landscape
On February 16, 2005 the Kyoto Protocol, the international agreement to address
climate disruption, went into effect and almost 200 countries have ratified it to date.
Despite the fact that the U.S. was a signatory to the agreement, Congress made it clear it
had no intention of ratifying the treaty, and thus the U.S. would not be participating in the
Kyoto Protocol (Regan, 2015; Layzer, 2006). Determined to address the issue, on that
same February day the Kyoto Protocol went into effect across the globe, Seattle Mayor
Greg Nickels launched the U.S. Mayors Climate Protection Agreement to advance the
goals of the Kyoto Protocol through local government leadership and action (US
Conference of Mayors, 2015).
There are more than 1,000 signatories to the agreement, which is now called the
US Conference of Mayors Climate Protection Agreement (USMCPA), including Denver,
Colorado (US Conference of Mayors, 2015). Participating cities have agreed to reduce
community-wide greenhouse gas (GHG) emissions to at least 7 percent below 1990
levels or better by 2012. The signatory cities have committed to meet or exceed the
Kyoto Protocol GHG emission targets in their own communities by using a litany of
climate-related polices and plans9. Climate and energy policy at the state level has also
9 For a review of the results of these trends see Krause, 2011 and Wood et al., 2014.


83
been created in response to the lack of national policy. At least 30 U.S. states, including
Colorado, and hundreds of U.S. cities, including Denver, have created a climate action
plan (CAP) of some sort (EPA, 2015). A city or state CAP typically outlines broad policy
goals as well as specific programmatic and policy recommendations that will be used as
greenhouse gas reduction measures.
Denver and Colorado Climate Policy Landscape
In 2005 then Denver Mayor John Hickenlooper launched Greenprint Denver, a new
department in the City and County of Denver created to advance and further support the
integration of environmental impact analysis into the citys programs and policies,
alongside economic and social analysis (City and County of Denver, 2006, p. 2). This new
program built on the citys history of climate policy going back a decade is an early
member of the International Council of Local Environmental Initiatives (now known
simply as ICLEI) (Bulkeley & Betsill, 2003). Compared to other medium to large U.S.
cities that are signatories to the USMCPA, Denver is typical in terms of its population
density, stated climate policy goals, and current progress in attaining said goals. In
November 2007 then Colorado Governor Bill Ritter launched an initiative to address
climate change statewide, which resulted in the creation of the Colorado Climate Action
Plan. This plan called for a reduction of the state emission of greenhouse gases by 20
percent below 2005 levels by 2020 and a longer-term goal of 80 percent below 2005 levels
by 2050 (Colorado Climate Action Plan: A Strategy to Address Global Warming, 2007). In
terms of scale and scope of policies, as well as the shorter-term emission reduction targets,
the Colorado CAP is typical of state CAPs (EPA, 2015; Ramseur, 2007). Given this,
Colorado could be thought of as a typical case in terms of case selection (Gerring, 2007).


84
Since the inception of both the plans, Mayor Hickenlooper, who spearheaded the
Denver level climate action plan, has become the Governor of Colorado and is now
leading the charge on the state CAP. The policy actors from across Colorado, including
those from Denver, will be treated together as subnational actors for the following
reasons: Denver is the largest city in Colorado, the states capital, and influential in state-
level policy developments in climate and energy issues, and often, as in this case, state
and city governments cooperate on energy and climate policy, lastly, given the leadership
and administrative changes at the local and state level, there is overlap in city and state
political actors.
DATA AND METHODS
Data Collection
This study is an investigation of the climate and energy policy subsystem of
Colorado using data collected in 2011 from an original internet-based survey. The target
population was policy actors employed by or involved with government agencies,
nonprofit organizations, and private companies engaged in climate and energy policy in
Colorado. This included individuals who attended the roundtable sessions preceding the
creation of Greenprint Denver and the Colorado CAP, and the advisory council members
of both plans. A snowball sample was generated from this set of individuals. The survey
was sent to 793 policy actors involved in climate and energy issues in Colorado and 260
individuals returned fully completed surveys (response rate of 33 percent). The sampling
technique employed for the survey is explicitly purposive and non-probability based
because the views and opinion of active policy actors is explicitly the goal of this study.


85
Data Analysis
The quantitative data collected with the surveys were analyzed using STATA 13
software. Descriptive statistics and crosstabs were calculated. Given the nature of the
dependent variable as non-ordinal scale of policy learning, multinomial logistic
regressions were used to test the effect of extreme beliefs and policy activities on belief
change or reinforcement. The normality, skewedness, and relevant assumptions regarding
the tests were examined, as well as the potential for any interactions between independent
variables.
Operational Measures
Table 3.1 below presents the operational measures used for each of the
independent variables and the dependent variable. In other words, the exact survey
questions used to measure each of the variables in this chapter, as well as the format of all
possible responses for each question, are presented below. For responses that were coded
for statistical purposes, that information is also provided for each question. For sake of
brevity, this is not a comprehensive list of all the survey questions.
Each variable concept is listed in bold, and below each concept is the operational
definition and survey question, used to measure each variable. Three different policy
activities measured the frequency with which the policy actors engaged in each activity
on a 5-point frequency scale. Extremeness of policy beliefs (independent variable for
hypothesis 2) was measured using a 5-point Likert scale of agree/disagree for six policy
belief statements. The dependent variable, individual policy learning, was measured for
each of the six policy belief questions (in other words, policy learning was measured six


86
times, once for each of the six policy beliefs), and six categories of responses were
offered regarding the form of policy learning.
Table 3.1. Operational Measures of Attributes
Policy Activities
Respondents were asked to report how often they participated in the following activities as part of their
work in the past year (on a scale of: daily, weekly, monthly, yearly, and never):
Collaboration with those who do not share their views on energy & climate goals
Seek advice from individuals they agree with
Facilitated consensus-building processes (e.g., focus groups, roundtables)
These were coded in the following way: Never = 0, Yearly = 1, Monthly = 2, Weekly = 3, Daily = 4
Policy Beliefs & Extremeness of Belief
Respondents were asked to indicate their relative agreement (on a scale of strongly agree, somewhat
agree, neither agree nor disagree, somewhat disagree, and strongly disagree) with the following
statements:
The severity of predicted impacts on society from climate change are vastly overstated
Human behavior is the principal cause of climate change
Decisions about energy and its effect on climate are best left to the economic market, and not to
the government
An energy and/or carbon tax is required to combat climate change
A cap and trade system of permits for the emission of greenhouse gasses is required to combat
climate change
Government policies to promote renewable energy generation are required to combat climate
change
Extremeness of Belief was created by coding answers for each policy belief question in the following way:
Strongly agree and strongly disagree = 2
Somewhat agree and somewhat disagree = 1
Neither agree nor disagree = 0
Operational Measures / Survey Questions: Learning
For each of the six policy belief questions above, respondents were asked to what extent their views had
changed or been reinforced (no time frame was associated with these questions):
Only been reinforced
Mostly reinforced with little changes
Balance of reinforcement and changes
Mostly changed with little reinforcement
Only been changed
Neither changed nor been reinforced


Full Text

PAGE 1

EXAMIN ATION OF POLICY AND POLITIC AL LEARNING: A STUDY OF COLORADO CLIMATE AND ENERGY POLICY ACTORS by ANDREW P ATTISON B.A., Skidmore College, 1999 M.P.A., University of Colorado Denver, 2007 A thesis submitted to the Faculty of the Graduate School of the University of Colorado in partial fulfillment of the requirements for the degree of Doctor of Philosophy Public Affairs Program 2015

PAGE 2

2015 ANDREW PATTISON ALL RIGHTS RESERVED

PAGE 3

! ii This thesis for the Doctor of Philosophy degree by Andrew Pattison has been approved for the Public Affairs Program by Paul Teske Chair Christopher Weible Advisor Tanya Heikkila Debbi Main Michele Betsill Date: October 9, 2015

PAGE 4

! iii Pattison, Andrew ( Ph.D., Public Affairs) Examination of Policy and Political Learning: A Study of Colorado Cl imate and Energy Policy Actors Thesis directed by Associate Professor Christopher Weible ABSTRACT The central goal of this dissertation is to examine factors that shape policy learning and political learning in policy actors. Policy learning is defined here as change and reinforcement in policy beliefs, and political learning is defined as change and r einforcement of advocacy strategy. The absence of either change or r einforcement was considered non learning ." Factors examined that may affect learning are "extreme" beliefs and various policy activities. The Advocacy Coalition Framework was used as a theoretical foundation to develop hypotheses. An original survey with 260 responding climate and energy policy actors in Colorado provided the data The results indicate extreme policy beliefs are associated with policy learning and political learning. Further, the product of that learning was more likely to be belief and advocacy strategy reinforcement as opposed to change Of the policy activities measured, increased collaboration with policy actors with d ifferent views showed some associat ion with mitigation of policy belief reinforcement Overall, these findings suggest some policy activities can stimulate policy learning and political learning and may serve as a balance to the reinforcement effects on policy and political learning of extreme beliefs The form and content of this abstract are approved. I recommend its publication. Approved: Christopher Weible

PAGE 5

! iv DEDICATION Dedicated to my wife and partner in all adventures, Chandra Russo. Without you and your support in my life this would, simply put, not have been possible. I do not have the capacity to thank you appropriately with words but will spend the rest of my life trying to do so with actions. I love you. Also to my brothers, and especially m y parents, for giving me a foundation on which to build my life, for inspiring me to always push myself further, and for alway s reminding where I came from.

PAGE 6

! v ACKNOWLEDGMENTS There is no way, zero, that I could have done this without the support and guidance of advisors and mentors in the world of academia Thank you first to my dissertation advisor Chris Weible for your suppor t and guidance in this process. Y ou continued to believe in me and push me long after I deserved your time and energy Your dedication to mentoring students and your pursuit of knowledge is an inspiration. Perhaps more than anything, thank you for showing me how brilliant and competent someone can be, while still being wise enough to be humbl e a model I will always aspire to equal and will likely stumble even coming close to. Thanks man. Thank you to my dissertation committee. Paul Teske, who walked over to my cubicle one day in 2006 and told me that I got into the Ph.D. program, that I would be supported by an NSF IGERT gra nt, and then explained what that heck that meant. Your skillful work as the Dean of a fantastic School of Public Affairs, your passion for "real world" politics, and most importantly the way you blend those two in your work in Denver was the re ason I deci ded to start this process in the first place Tanya Heikkila, who was willing to join my committee when I needed her and provided invaluable insight and inspiration helping me to thin k deeply and write simply Your professionalism, energy and humor came to SPA at the perfect time for me and you made me believe again that this was something I could and wanted to do. Debbi Main y our passion for community based participatory research helped me see that our work in Sustainable Urban Infrastructure ca n do so much more than improv e theories, it can also improve lives. Michele Betsil l I cited your book on cities and climate change in my application to the Ph.D. program and your CV read like how I wanted mine to. I never

PAGE 7

! vi dreamt in 2006 that your wisdom and experience would one day be put to improving my writing, methods, and arguments. Thank you all This dissertation is better because of your efforts, and any remaining shortcoming s are all mine Thank you to other teachers along the way like Linda d e Le on, Christine Martell, Lloyd Burton, and Anu Ramaswami. Thank you especially to George Busenberg and Peter de Leon, my first two M.P. A. class professors, for opening my eyes. You a re all my inspiration. T hank you to all of my cohort members. T hough we have scattered to the winds over the years, we pushed and pulled each other through those first years with hard work and consoled each other with laughs Thank you especially to Saba Siddiki, John Calanni Scott Mendelsberg and Laurie Mandrin o. Thank you to Dallas Elgin for agreeing to team up on the survey, for advice along the way, and for modeling professionalism at every point. I am incredibly grateful to Paul Teske and Anu Ramaswami for the opportunity to have been supported by the NSF IGERT grant 2007 2 010. That sustainable urban infrastructure program provided incredible financial support and invaluable learning opportunities. I feel even more blessed and privileged to have been part of that program with every year that passes. Thank you to Chris Weible Tanya Heikkila and Debbi Main for continuing to lead tha t program in the later years. Thank you to my good friend s outside the world of UCD SPA. I am blessed to have too many to name them all here, but you know who you are. For teaching me so much as we hike, camp, ski, swim, ride, debate, play, laugh, and fish our way through this fragile and miraculous life. Nothing worth having in life is easy to get, including an amazing group of friends.

PAGE 8

! vii Thank you to my family, and specifically my parents Art and Kris, for providing me every opportunity to learn and grow, for having patience with me in the circuitous path I have taken in all of my employment choices, and for all the support to become my own person as I did so. Thank you to the Pattison clan, from the first member of the family to go to college, to a Ph.D. in only two generations not too shabby. Thank you to the St ern clan, with four Ph.Ds. in 14 grandkids we are so privileged, and so ridiculous. Finally and most importantly, thank you to Chandra D urin g the time I worked on this you were more supportive, empathetic, and u nderstanding than I can believe possible. Yo u are the rock that weathered all aspects of my insanity and anxiety during this process. Sorry about that. As always, I look forward to the next adventure with you.

PAGE 9

! viii TABLE OF CONTENTS CHAPTER I. INTRODUCTION ................................ ................................ ................................ .. 1 Theory: The Advocacy C oalition Framework and Learning ................................ 4 Dissertation Objectives, Research Questions and Hypotheses by Chapter ................................ ................................ ................................ ............ 14 The Case: Colorado Climate and Energy Policy Subsystem ............................... 17 Methods: Data Collection and Analysis ................................ ............................... 21 Contributions ................................ ................................ ................................ ........ 23 I I SUMMARY ANALYSIS OF CLIMATE AND ENERGY POLICY ACTORS IN COLORADO ................................ ................................ .. 28 Chapter Summary ................................ ................................ ................................ 28 Introduction ................................ ................................ ................................ .......... 29 The Advocacy Coalition Framework ................................ ................................ ... 32 Policy Actor Attributes Measured ................................ ................................ ........ 35 Climate and Energy Policy in Colorado ................................ ............................... 42 Methodology ................................ ................................ ................................ ........ 44 Results ................................ ................................ ................................ .................. 48 Conclusions ................................ ................................ ................................ .......... 65 II I. FACTORS SHAPING POLICY LEARNING: A STUDY OF POLICY ACTORS IN SUBNATIONAL CLIMATE AND ENERGY ISSUES ................................ ................................ ............................... 70 Chapter Summary ................................ ................................ ................................ 70 Introduction ................................ ................................ ................................ .......... 71 Theory & H ypotheses: Policy Learning and t he Advocacy Coalition Framework ................................ ................................ ............................ 73

PAGE 10

! ix Climate Change Policy ................................ ................................ ......................... 81 Data and Methods ................................ ................................ ................................ 84 Results and Analysis ................................ ................................ ............................ 87 Discussion ................................ ................................ ................................ .......... 104 Conclusion ................................ ................................ ................................ .......... 108 IV. FACTORS SHAPING POLITICAL LEARNING: A STUDY OF POLICY ACTORS IN SUBNATIONAL CLIMATE AND ENERGY ISSUES ................................ ................................ ............................. 113 Chapter Summary ................................ ................................ ............................... 113 Introduction ................................ ................................ ................................ ........ 114 Theory and Hypotheses: Po litical Learning and t he Advocacy Coalition Framework ................................ ................................ .......................... 117 Climate Change Policy ................................ ................................ ....................... 126 Data and Methods ................................ ................................ ............................... 130 Results and Analysis ................................ ................................ .......................... 133 Discussion ................................ ................................ ................................ .......... 140 Conclusion ................................ ................................ ................................ .......... 145 V. CONCLUSION ................................ ................................ ................................ .. 148 Dissertation Summary ................................ ................................ ........................ 148 Chapter Two Objectives and Summary Findings ................................ ............... 149 Chapter Three Research Question, Hypotheses and Summary Findings ................................ ................................ ................................ .............. 149 Chapter Four Research Question, Hypotheses and Summary Findings ................................ ................................ ................................ .............. 150 Synthesized Findings and Discussion ................................ ................................ 151

PAGE 11

! x Limitations ................................ ................................ ................................ .......... 156 Contributions ................................ ................................ ................................ ...... 158 Future Research ................................ ................................ ................................ .. 166 R EFERENCES ................................ ................................ ................................ ................ 169 APPENDIX A. Output from Chapter Three Multinomial Logit Regression ............................... 181 B. Output from Chapter Four Multinomial Logit Regression ................................ 187

PAGE 12

! xi LIST OF TABLES TABLE 2.1. Operational Measures of Attributes ................................ ................................ ....... 45 2.2 Percentage of Respondents That Received Training in Applied Research ................................ ................................ ................................ ................. 56 2.3. Respondent Information Use Summary ................................ ................................ 59 2.4. Political Learning in Survey Respondents ................................ ............................. 65 3.1. Operational Measures of Attributes ................................ ................................ ....... 86 3.2. Extremeness of Policy Beliefs ................................ ................................ ............... 87 3.3. Policy Learning Across the Six Policy Belief Questions ................................ ....... 89 3.4. Independent Variable Labels in Marginal Change Plots ................................ ....... 93 5.1 Summary of Support for Chapter Three Hypotheses ................................ ........... 150 5.2. Summary of Support for Chapter Four Hypotheses ................................ ............ 151

PAGE 13

! xii LIST OF FIGURES FIGURE 2.1. Survey Respondents by Political Orientation ................................ ........................ 49 2.2. Organizational Affiliation of Survey Respondents ................................ ................ 50 2.3. Respondents by Formal Education ................................ ................................ ........ 51 2.4. Respondents by Years Participati ng in Climate and Energy Issues ...................... 52 2.5. Respondents by Frequency of Engagement in Policy Activities ........................... 53 2.6. Extent of Policy Capacity in Respondents' Organization ................................ ...... 55 2.7. Priority of Climate and Energy Related Issues ................................ ...................... 56 2.8. Respondents Reported Access to Professional R esearch ................................ ....... 57 2.9. Policy Beliefs ................................ ................................ ................................ ......... 61 2.10. Policy Learning Across the Six Policy Belief Questions ................................ ....... 64 3.1. Frequency of Policy Activities ................................ ................................ ............... 88 3.2. Changes in Predicted Probabilities in Learning Regarding the Severity of Predicted Impacts of Climate Change ................................ ................. 94 3.3. Changes in Predicted Probabilities in Learning Regarding Humans as the Cause of Climate Change ................................ ................................ ............ 95 3.4. Changes in Predicted Probabilities in Learning Regarding the Need for Government to Address Climate Change ................................ ......................... 96 3.5. Changes in Predicted P robabilities in Learning Regarding the Need for a Carbon Tax to Address Climate Change ................................ ....................... 97 3.6. Changes in Predicted Probabilities in Learning Regarding the Need for a Cap & Trade System ................................ ................................ ..................... 98 3.7. Changes in Predicted Probabilities in Learning Regarding the Need for Policies Promoting Renewables ................................ ................................ ....... 99 4.1. Policy Actor Extremeness of Belief ................................ ................................ ..... 134

PAGE 14

! xiii 4.2. Learning Regarding Advocacy Strategies ................................ ........................... 135 4.3. Changes in Predicted Probabilities in Learning Regarding Advocacy Strategy for Climate and Energy Policy Issues ................................ .. 138

PAGE 15

! 1 CHAPTER I INTRODUCTION Nested within the broader field of public affairs is a subset of public policy research aimed at understanding what has come to be called the policy process. Harold Lasswell beginning in the 1940s and 19 50s articulated one of the first theoretical frameworks used to understand this new field using conceptual maps and a model of generalizable decision making processes he called the policy sciences (de Leon 2006, p. 39). T he modern study of policy process research is defined as the examination of interactions between public policy and the relevant individuals, events, setting s and outcomes of policies (Weible, 2014b, p. 5) The study of these interactions can now be approached from a diverse set of policy process of theories, m odels and frameworks (Cairney & Heikkila 2014) The Advocacy Coalition Framework (ACF) is one of these approache s The ACF conceives of policymaking as the interaction of competing values of the opinion leaders and policy implementers working on a particular policy problem These specialized individuals, often referr ed to as policy actors ," have invested time and resources gaining knowledge of specific policy area s and buil ding the relevant political networks ( e.g. advoca cy coalition s ) to better help affect the development of public policy ( Jenkins Smith et al., 2014 ; Kraft & Furlong, 2015; Birkland, 2011 ). Conflicts can emerge whe n there are a diversity of opinions and policy positions held by policy actors and/or coalitions across a variety of factors These might include : the severity and primary cause of environmental problems, the need for the role of the government in ecosystem protection and the relative prioritization of specific policy preferences As

PAGE 16

! 2 policy actors engage and interact newly a cquired i nformation and experience s are interpreted in ways that may change or reinforce the beliefs and behaviors of the individual and coalitions engaged in the policy process. Th ese change s in belief and behavior over time ha ve been d efined as learning ." T he concept of l earning has long played a central role in the ACF and other theories and frameworks used to understand policy processes (Lindblom, 1956; Heclo, 1974; Be nnett & Howlett, 1992 ; May, 1992; S abatier & Jenkins Smith, 1993; Gerlak & Heikkila 2013 ) This study will differentiate between policy learning, pertaining to cognitive change and /or reinforcement in policy beliefs, and political learning, regarding behavioral change and /or r einf orcements in advocacy strategy employed in policymaking. Policy actor learning, defined as b elief and behavior change or belief and behavior reinforc ement may be shaped by factors such as cognitive abilities, perceptual filters, or existing belief structures ( Lord et al ., 1979; Simon, 1985). Other factors affecting learning may be experience s such as participation in policy activities between ally and opponent coalitions that advocate diff erent policy goals ( Sabatier & Jenkins Smith 1993 1999) For example, the extent to which policy actors engage in policy activities such as collaboration an d consensus building with individuals or coalitions with dissimilar policy views and goals may affect information acquisition and potentially shape learning ( Sabatier & Jenkins Smith 1993 1999 ; Jenkins Smith et al., 2014 ) This dissertation asks and seeks to answer the following general question: What factors influence how information and exp eriences shape policy actor learning ? S ome factors may lead to different forms of learning. For instance, some factors may shape learning as changes in beliefs and behavior and some factors may reinforce beliefs and behavior. Acquiring a

PAGE 17

! 3 better understandi ng of these factors m ay help to build more effective policy processes, processes better equipped to navigate what might appear as intransigent conflicts between indi viduals and groups working on issues suc h as climate and energy policy. The field of policy process theory is founded on the observation that public policy does not develop within a single government institution, but by subunits of a political system in what have been called policy subsystems (Redford, 1969). Policy subsyste ms continue to be a primary unit of analysis in the field of policy process as it has been evident from decades of research that policies often are created and evolve through the work of multiple overlapping involved organizations, groups, and coalitions o f specialized policy actors ( Birkland 2011; Sabatier & Jenkins Smith 1993). Using a study of the Colorado climate and energy policy subsystem and the ACF as a lens through which to view the attributes, beliefs, and learning of the policy actors therein the major objectives of this dissertation are to: 1) M easure and understand the diversity of various att ributes of these policy actors and the extent of policy actor policy and political learning in this subsystem 2) E xamine what if any effect s extreme beliefs and policy activities have on policy actor policy learning 3) Examine what if any effects extreme beliefs and policy activities have on policy actor political learning. THEORY: THE ADVOCACY COALITION FRAMEWORK AND LEARNING The Advocacy Coalition Framework (ACF), developed by Sabatier and Jenkins Smith in the late 1980s, evolved from a long history of research on environmental problems, and was an attempt to reconcile top down and bottom up theories of policy

PAGE 18

! 4 implementation (Sabatier 1988 1998 ; Jenkins Smith et al ., 2014 ). Throughout its development the ACF has become a leading framework to examine and explain long term policy change individual policy actor belief and behavior, and change s in beliefs and strategies ( Jenkins Smith e t al., 2014 ). As with other social science theories and frameworks the ACF builds from the social psychology model of the boundedly rational individual possessing perceptual limits and filters stemming from preexisting belief structures ( Sabatier 1988; S abatier & Jenkins Smith 1993 ). T he primary unit of analysis assumed by the ACF is the policy subsystem. According to the ACF, two dimensions characterize a policy subsystem: a substantive dimension and a territorial dimension ( Jenkins Smith et al ., 2014 ). In this study the substantive dimension is climate and energy policy ( specifically in this case regarding policy pertaining to climate change, greenhouse gas mitigation, or climate action in general) and the territorial dimension is Colorado (meaning policy actors who are active within the State of Colorado, and Colorado cities, for example the City and County of Denver ). Research under the ACF has focused on a variety of phenomenon, such as the extent that policy actors are motivated by and filter newly acquired information through their beliefs while attempting to influence policymaking The logic of such a pursuit comes from an assertion that individual beliefs are the causal driver for political behavior and pull like minded actor s together in advocacy coalitions ( Sabatier, 1988 ) The ACF defines advocacy coalitions as group s of legislators, agency officials, interest group leaders, and researchers with similar policy beliefs that share resources and engage in a nontrivial degre e o f coordination ( Jenkins Smith et al., 2014 ).

PAGE 19

! 5 Policy actors in the climate and energy policy subsystem are expected to have a variety of beliefs regarding ca u sal mechanisms such as the relative contribution to natural global climate change caused by human behavior and specific policy preferences like the need for government intervention into the energy market Belief and value clashes within the policy process are likely involved in the lack of substantive national climate policy, but policy change has been occurring at the local level over time despite goal conflicts and technical disputes. Potentially, this indicates learning is occurring in local energy and climate policy subsystems, thus indicating an appropriate case to use the ACF to test some hyp otheses related to policy actor learning Theoretical Emphasis The ACF puts forth a number of causal relationships, which serve as a theoretical foundation in this study (for a full list of ACF hypotheses see: Jenkins Smith et al ., 2014 ). There are three m ajor ACF theoretical emphases which applications in the literature typically focus upon : 1) advocacy coalitions ( understanding and explaining coalition structure, membership, resources, and changes over time ); 2) learning or policy oriented learning" (changes in policy actor beliefs or the use of political strategies for achieving objectives ); and 3) policy change ( changes in core aspects of governmental policies or the beginning or end of specific programs) (Jenkins Smith et al., 2014) This study wil l focus on theories regarding learning and will specifically examine both policy learning (pertaining to changes and reinforcement in policy beliefs) and political learning (pertaining to behavioral changes and reinforcements in advocacy strateg y ). To cont extualize this study's examination of policy and political learning, some discussion of the way learning i s studied in disciplines other than public affairs is necessary.

PAGE 20

! 6 V arious social and natural science fields have struggled to understand learning for decades (Grin & Loeber, 2007; Milner, Squire & Kandel, 1998) Despite a great deal of scholarship in regards to this topic, it is difficult to make comparisons across much of the work. Muro and Jeffery (2008, p. 327) claim that this difficulty is due to th e " different underlying assumptions about the na ture of learning of knowledge " In other words different conceptualizations of learning within and across disciplines ha ve confounded attempts to make synthesizing theories and models of individual learni ng (Reed et al ., 2010). Recent meta stu dies of learning (see Murro & Jerry 2008; Reed et al ., 2010; Crona & Parker 2012) agree that while no singular definition, conception, or operationalization of learning across the social sciences exists, learning is often cited as a normative goal and perhaps a prerequisite for collaborative and adaptive governance and sustainable policymaking. It is there fore important to delineate different dimensions of learning. As in other disciplines, the investigation of lear ning in the policy process literature has a long and eclectic history ( Heikkila & Gerlak 2013; Freeman 2006; Grin & Loeber 2007). Heclo's (1974, p p 303 306) discussion of learning as "collective puzzling" as a factor in policy change is often cited in public policy literature and influenced Sabatier's early work in developing the ACF. Bennett and Howlett's (1992) effort to synthesize theories of learning in public affairs is useful to unpack some of the various dimensions of learning and may allow for l arger lessons to be drawn from the complicated field of learning. Bennett and Howlett (1992) describe a framework to describe learning studies across three dimensions : 1) who is learning, 2) what is being learned, and 3) the results of learning. For exampl e, individual policy actors may learn about the efficacy of certain climate and energy policies lead ing to policy change over time.

PAGE 21

! 7 In terms of who is learning ," distinctions must be made between individual learning and group learning (Reed et al ., 2010) Mixing methodologies or concepts across different units of analyses can lead to difficulties. This dissertation is examining individual learning to better understand the factors that shape policy actor learning in this policy subsystem. In terms of what is being learned ," individuals can learn about any aspects of public policies. May (1992) provides a useful distinction between social policy learning (the social construction of the policy problem or goal for example), instrumental policy learning (such as the efficacy of various policy tools or designs), and finally political learning (which would involve knowledge of the viability or public palatability of various policy tools or instruments). This dissertation explores both policy learning and politica l learning Policy learning (encompassing both s ocial policy learning and instrumental policy learning ) will be examined in the form of change and reinforcement of policy actors' policy beliefs P olitical learning will be measured in the form of policy act ors' advocacy str ategy change and reinforcement. The typologies of learning de fined by Bennett and Howlett ( 1992 ) and May (1992) helped organize the public affairs literature initially and recent studies are attempting to investigate further and explore the role of learning in the policy process (Heikkila & Gerlak, 2013; Gerlak & Heikkila 2011) A crucial differentia tion is that the results of learning" (such as belief change or reinforcement of advocacy strategies) are distinct from the process of learning itself meaning the factors that shape the learning such as various sources of information or policy activities The results or outcomes of learning may be conceptualized in numerous ways, for instance as policy change (Busenberg 2001). Some res earchers make distinctions between cognitive versus

PAGE 22

! 8 behavioral changes (Gerlak & Heikkila 2011) while others look for organizational changes in collaborative settings (Crona & Parker 2012). The contextual factors that describe how social environments an d social networks cultivate learning also seem important to delineate ( Heikkila & Gerlak 2013 ; Reed et al ., 2010), as do broader descriptions of the po litical environment (Gerlak & Heikkila 2011). Learning, as a part of the policy process, is particularl y relevant within the ACF ACF scholars have explicitly argued since its in ception that in the attempts by various policy actors to affect policies and programs efforts are made to improve their understanding of the problem and alternatives, thus leading to individual learning (Sabatier, 1988) 1 Within the ACF, learning has been defined as continuing adjustments to beliefs and behaviors concerning policy objectives resulting from new experiences or information (Sabatier & Jenkins Smith 1999, p. 123). However learning has been inconsistently and /or poorly defined both theoretically and operationally across ACF applications ( Weible et al., 2011). M any applications of ACF have affirmed some of the underlying theoretical claims For instance, that opposin g advocacy coalitions within policy subsystems are formed and maintained based on stable policy beliefs and t hat policy change is associated with ma jor policy or political events (Jenkins Smith et al., 2014). L earning however has been the least explored of the three major ACF theoretical emphases (Jenkins Smith et al ., 2014; Weible et al ., 2009; Zafonte & Sabatier 1998 ) In the studies conducted over the previous decades scholars applying the ACF have found learning to be more likely in a context where : 1) conflict is at intermediate !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! !!!!!!!!!!!!!!! 1 In the introduction of the ACF Sabatier used the term policy oriented learning ." T his study will follow the lead of other scholars employing the ACF and abbreviate this as learning ." !

PAGE 23

! 9 levels and focuse d on beliefs related to specific policy design alternatives, as opposed to more fundamental beliefs such as causal arguments or political orientation ; 2) issues in conflict are technical in nature (as opposed to social ) ; and 3) when there are professional forums for individuals in opposing coalitions to collaborate on policy development ( Lester & Hamilton, 1987; Sabatier & Brasher, 1993; Eberg 1997; Ellison 1998; Lubell, 2003; Meijerink, 2005; Larsen et al., 2006). A 2009 (Weible et al ) comprehensive meta analysis of applications of the ACF found that some studies have shown policy learning and individual belief change within and across coalitions to have potential to drive policy development and chan ge over time. But multiple subsequent surveys of ACF theory and applications argue that one major area within the ACF needing innovation and improved clarity in theory and model building as well as methodology is learning ( Jenkins Smith et al ., 2014; Weible et al ., 2011 ). By exploring both policy learning and political learning as distinct concepts this dissertation aspires to contribute to the improvement of the study of learning in the policy process. Using the ACF to Study Learning One of th e essential challenges in studying learning, in general and using the ACF, is how learning is defined conceptually. For example is policy actor belief change the only necessary and sufficient indicator of learning? Or is coalition change or is policy change that indicator? In the ACF literature, one or more of these three "indicators" of learning are often provided as evidence of learning but rarely is this explicitly linked to a specific conceptualization of learning. Even fewer ACF applicatio ns have attempted to measure learning quantitatively using survey data; instead, researchers

PAGE 24

! 10 have typically relied on unsystematic assessments of policy chang e or content analysis (Weible et al., 2011) O ther potential indicator s of learning posited by the ACF are belief reinforcement, advocacy strategy change, and advocacy strategy reinforcement all of which may occur based on different learning processes of policy actors For instance, some experiences in the policy process may serve to change or refine policy actors' beliefs and behaviors, while participation in different policy activities may serve to strengthen and galvanize beliefs and behaviors (Heikkila & Gerlak 2013) Relatedly, some attributes such as extremeness of preexisting beliefs may fact or into the learning process of policy actors To further narrow a conceptual definition of learning, the various dimensions of the learning concept from Bennett and Howlett (1992) and May (1992) are again helpful. To begi n, in this study, the who is learning" will be policy actors in the climate and energy subsystem of Colorado. I ndividual policy actors will be the unit of analysis As to the what is being learned ," this study focuses on two aspects of learning. C hapter T hree examines policy learning ( or social and instrumental policy learning using May's 1992 typology) by examining cognitive belief change and belief reinforcement, as well as the absence of either, regarding a range of policy information across the energy and climate policy subsystem. Specifically, policy beliefs related to causal mechanisms and the perceived severity of the problem of climate change will be examined. Additionally, beliefs examined regarding the r ole the government (versus the economic market) should play in addressing the problem as well as beliefs related to the perceived need for various specific policy solutions. This kind of learning r elated to the social construction of the problem and learning regarding the viability of specific policy

PAGE 25

! 11 implementation designs a re exemplars of social policy learning and instrumental policy learning respect ively in May's (1992) typology. Chapter Four explores political learning by measuring behavioral change and reinforcement of advocacy strategies, as well as the absence of eithe r. Thus policy actors' beliefs and behaviors are i n cluded within the dimension of what is being learned" in this study and thus span across all three of May's (1992) typology of policy and political learning dimensions Other fields such as sociology recognize and study the role of political actors' strategy responses based on changing political opportunities and threats in policymaking processes, cont entious or otherwise (Tilly & Tarrow 2007). The ACF postulates that coalition and individual actors' advocacy strategies may shift based on, for instance changes in short term political constraints or resources (Jenkins Smith et al ., 2014). This c hanging or reinforcing of advocacy strategy based on awareness of political feasibility is an example of Ma y's (1992) political learning. Within the context of subnational climate and energy policy actors specifically, actors may adjust advocacy strategies based on political factors regarding the perceived need for local governments to address climate change or sociotechnical factors such as trends in energy production and infrastructure construction (Bulkeley 2013). T he direct examination of policy actors' political learning regarding advocacy strategies is underdeveloped compared to issues related to coalitio n resources and activities more broadly (Jenkins Smith et al ., 2014). Usin g Bennett and Howlett (1992) ty pology, this study defines policy actors report s of their cognitive and behavioral change s (or reinforcement s ) as the results of learning ." These cognitive and behavioral reinforcement and changes could also be

PAGE 26

! 12 categorized as what Heikkila and Gerlak (2013) would refer to as the products of learning. T his dissertation seeks to specifically differentiate between belief change and belief reinfo rcement as well as change and reinforcement of advocacy strategies as different products of policy learning and political learning respectively While change and reinforcement (cognitively or behaviorally) may be included in a concept of learning in the ACF, this study argues they are different categories of learning with different, and perhaps different implications. U nderstanding the factors that shape learning, the processes of learning resulting in differ ent products, and delineating between change a nd reinforcement is the explicit goal of this study Chapter Three of this study specifically conceptualizes and defines belief reinforcement in addition to belief change, as evidence of conative or policy learning Similarly, Chapter Four conceptualizes and defines reinforcement in addition to change, in policy actor advocacy strategies as evidence of behavioral or political learning Another challenge in st udying the concept of learning using the ACF, or any set of policy process theories or frameworks, is understanding what Goertz (2006 p. 178) would refer to as "the negative pole" or the "negative concept." Specifically, if cognitive and/or behavioral change as well as reinforcement can constitute learning, what constitutes non learning "? This quest ion is rarely considered in studies of learning across many fields of study (Heikkila & Gerlak 2013). E vidence of non learning could be, for example, the absence of belief change or policy change. Interpretations of early applications of policy learning t heories of the ACF would seem to indicate that since policy change is prima fascia evidence of learning, then no policy change is e vidence of non learning. But that logic rests on the assumption that policy change indicates and

PAGE 27

! 13 comprises the concept of lea rning and that may not always be true. In science, when measuring a phenomenon it is preferable to measure variables across categories of positive and negative states. While scale data would be ideal, it is not possible with all conceptualizations of lea rning. W ithin the ACF learning is defined as: "relatively enduring alterations of thought or behavioral intentions that result from experiences and/or new information and that are concerned with the attainment or revision of policy objectives" (Sabatier & Jenkins Smith 1999, p. 123). In this case, "alter ations includes both change and reinforcement in beliefs and behaviors Thus, absence of change or reinforcement (in either policy beliefs on in political behaviors) will be evidence for not learning ." I n Chapter Three policy learning will be measured as self reported changes or rein forcement of beliefs across six b elief structures. Chapter Four will examine political learning as self reported change or reinforcement of advocacy strategies. Individuals a re asked directly if their beliefs and strategies have been changed or reinforced, as well as what the source of that change or reinforcement was This dissertation is one of the first ACF applications to specifically ask policy actors to report learning as change and reinforcement, measure both as well as the absence of either, and ask about the source of that learning Multinomial logistic regression analyses were u sed to examine the role of different factors in shaping different policy learning and political learning products This dissertation aims to explore, test and refine models of individual policy actor learning specifically to examine some of the factors that shape policy learning and political learning in this policy subsystem. In employing the ACF all attempts have been

PAGE 28

! 14 made to clarify assumptions, methods and approach. This dissertation is written as thr ee stand alone chapters with an introduction and conclusion. This study begins by providing an overall descriptive analysis of various attributes of the policy actor survey respondents including but not limited to the distribut ion of variation in beliefs, behaviors and learning The second and third chapters examine factors shaping policy learning and political learning respectively withi n this sample of policy actors. DISSERTATION OBJECTIVES, RESEARCH QUESTIONS, AND HYPOTHESES BY CHAPTER Chapter Two: Summary Analysis of Climate and Energy Policy Actors in Colorado Chapter two provides an overall analysis of the sample of policy actors, including demographics for instance, that responded to the survey. Descriptive statistics are provided for a host of variables measured in the sample including the extent of policy learning and political learning occurring in this subsystem T he two main objective s guide this chapter to: 1) U nderstand and describe attributes of the policy actors ( individuals in the government, private, nonprofit, and academic sectors) involved in climate and energy policy debates in Colorado including their beliefs across a number of relevant policy questions and 2) To determine the extent of policy learning, as defined as the range between belief change and belief reinforcement, and the extent of political learning, as defined by a change or reinforcement of advocacy strategies, in the sampled population of policy actors.

PAGE 29

! 15 Chapter Three: Factors Shaping Policy Learning: A Study of Policy Actors in Subnational Climate and Energy Issues Chapter T hree examines policy learning, as measured by belief changes and reinforcement, in the sample set of policy actors that replied to the survey. Various factors are measured to determine the effect on policy learning. Research Question : What factors affect policy actor belief change and reinforcement in a local and state level energy and climate policy subsystem? In order to explore the effect of these factors on policy learning, the followin g four hypotheses are presented an d tested for using survey data. Extreme Belief Hypothes i s : Hypothesis 1: Policy actors with more extreme policy views are more likely to reinforce their beliefs than change their beliefs. Policy Activities Hypotheses : Hypothesis 2 : Policy actors that seek advice more frequently from those with similar beliefs are more likely to reinforce their beliefs, rather than change their beliefs. Hypothesis 3: Policy actors that collaborat e more frequently with those with dissimil ar beliefs are more likely to change their beliefs, rather than reinforce their beliefs Hypothesis 4 : Policy actors that have participated in more frequent facilitated consensus based processes are more likely to change their beliefs, rather than reinforce their beliefs

PAGE 30

! 16 Chapter Four : Factors Shaping Political Learning: A Study of Policy Actors in Subnational Climate and Energy Issues Chapter F our examines political learning, as measured by change and reinforcement of advocacy strategies, of the policy actor respondents to the survey. Different variables are measured to determine the effect on political learning. Research Question : What factors affect policy actor political learning in a local and state level energy and climate policy subsystem? With the goal of exploring the effect of these factors on political learning, three hypotheses were examined using da ta from the survey respondents. Extreme Belief Hypothesis : Hypotheses 1: Policy actors with extreme beliefs will be more likely to change o r reinforce advocacy strategies than not change or reinforce advocacy strategies. Policy Activities Hypotheses : Hypothesis 2: Policy actors that participate in coalition building will be more likely to change or reinforce advocacy strategies than policy actors that do not parti cipate in coalition building. Hypothesis 3: Policy actors that participate in facilitated multi stakeholder consensus based processes will be more likely to change or reinforce advocacy strategies than policy actors that do not part icipate in negotiated multi stakeholder consensus based processes. Hypothesis 4: Policy actors that participate in coalition building will be more likely to report advocacy strategy change than advocacy strategy reinforcement.

PAGE 31

! 17 Hypothesis 5: Policy actors t hat participate in facilitated multi stakeholder consensus based processes will be more likely to report advocacy strategy change than advocacy strategy reinforcement. THE CASE: COLORADO CLIMATE AND ENERGY POLICY SUBSYSTEM The findings of the Intergovernmental Panel on Climate Change and the broader scientific community demonstrate that climate change could have dramatic impacts on human communiti es in the near future (Giddons 2011; IPCC 2013 2014 ; Stern 2007). Despite this threat the U.S. has yet to produce meaningful federal climate legislation 2 This has provided the opportunity for climate policy innovation at the subnational level Cities, towns, counties and states have creat ed and implemented a litany of climate and energy programs and legislation in the U.S. Thus, the study of these subnational climate policy subsystem case studies has beco me i mportant among political scientists and scholars struggling to understand the role of policy actors and the polic y process Previous research, specifically regarding subnational climate policy, h as led to insights regarding the need to further examine the actions of policy makers (Davis & Weible 2011), the drivers of local and state level decisions to commit to clim ate protection (Rabe 2004; Fogel 2007 ; Selin & VanDerveer 2007; Zahran et al. 2008 ; Krause, 2011), and issues of local climate governance (Betsill 2001; Bulkeley & Betsill 2003 ; Bulkeley 2013 ). This progress in the study of the politics and policies of subnational units has evolved alongside efforts in the engineering and other physical sciences. Pioneering work related to the creation of standardized methods for greenhouse !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! !!!!!!!!!!!!!!! 2 At the time of writing, President Obama is attempting to advance more substantive climate policies at the national level, but political constraints on implementation ha ve continued to slow progress.

PAGE 32

! 18 gas accounting which was done at the subnational level, and especially city level, focused on: modeling urban carbon flows with ecosystem theory (Churkina 2008), urban demand centered approaches to differentiate cross political jurisdictional and point of use emissions (Hillman & Ramasw ami 2010; Ramaswami et al. 2008 ), and the relationship between carbon emissions and urban development (Glaeser & Kahn 2010). Progress in the development of meta theories to understand and improve social ecological systems management and subnational climate policy, specifically, will be an interdi sciplinary effort (Ramaswami et al. 2012b). This study strives to be one aspect of that effort in further developing models of individual learning of policy actors and will employ a single policy subsystem study design ed to explore factors that shape lear ning, by applying the theoretical lenses of the ACF to the clim ate and energy policy subsystem of Colorado. Subnational Climate Policy Landscape The Kyoto Protocol, the international agreement to address climate disruptio n with changes to energy policies went into effect in 2005 and almost 200 countries have ratified it to date. While the U.S. is a signatory to the Kyoto Protocol Congress has continually refused to formally ratify the treaty, preventing full U.S. participation ( Regan, 2015; Layzer 200 6) 3 In an effort to advance the goals of the Kyoto Protocol through local government leadership and action on the same day the treaty went into effect across the globe, then Seattle Mayor Greg Nickels launched the U S Mayors Climate Protection Agreement (USMCPA) (US Conference of Mayors, 2015 ). Currently, t here are more than !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! !!!!!!!!!!!!!!! 3 At the time of writing, the Obama Administration is attempting to negotiate a more comprehensive international climate treaty. Given the complexities and uncertainties of international agreements, and the need for Congressional ratification of U.S. treaties, the details of any eventual policy ou tcome are currently unknowable.

PAGE 33

! 19 1,000 signatories to the USMCPA including Denver, Colora do (US Conference of Mayors 2015 ). Participating cities have agreed to reduce community wide greenhouse gas (GHG) emissions by 2012 to at least 7 percent below 1990 levels or better in their own communities by using a litany of climate related polices and plans Climate and energ y p olicy at the state level has also been created. Early climate and energy policy innovations at the state level were categorized and examined by Rabe (2004) W hile many of the policies were symbolic at first, later and especially recent developments ha ve become much more aggressive in terms of the stated goal s and the policy detail s ( Bulkel e y, 2013; Krause 201 1; Ramseur 2007). At least 30 U.S. states, including Colorado, in addition to hundreds of U.S cities have now created a climate action plan (CAP) of some sort ( EPA 201 5 ). A CAP typically outlines policy goals and recommendations that a state or city will employ to address climate change by making specific policy actions toward reducing the GHG emissions of that entity. Denver and Colorado Climate Policy Landscape Then Mayor John Hickenlooper launched Greenprint Denver in 2005. This new City and County of Denver department was created to "advance and further support the integration of environmental impact analysis into the city's programs and polici es, alongside economic and social analysis" (City and County of Denver 2006 p. 2). The program built on the history of Denver climate and energy policy going back a decade as an early member of the International Council of Local Environmental Initiatives (now know simply as ICLEI) (B ulkeley & Betsill 2003). The Greenprint Denver Plan includes a 10 point action agenda to reduce the city's environmental i mpact. It is a comprehensive list covering a variety of environmental policies. The research study proposed here will

PAGE 34

! 20 concern those action items dedicated to climate change policy and action. The first and perhaps most ambitious action agenda item from the Greenprint Denver Plan is to reduce Denver per capita greenhouse gas emissions by 10% below 1990 levels by 2011. Work in partnership with regional mayors, universities, and the business community to develop and implement effective strategies for adaptati on to and amelioration of global climate change" (City and County of Denver 2006 p. 6). Compared to other medium to large U S cities that are signatories to the USMCPA, Denver is typical in terms of its population density, stated climate policy goals, a nd current progress in attaining said goals. In November 2007 t hen Colorado Governor Bill Ritter launched an initiative to address climate change statewide, which resulted in the creation of the Colorado Clim ate Action Plan Th e plan called for a reduction of the state emission of g reenhouse gases by 20 percent by 2020 and was created in a collaborative manner from a diverse set of stakeholders " including business and community leaders, conservationists, scientists and concerned citizens" (Ritte r 2007, p. 2). The Colorado CAP is similar to the approximately 30 other state plans in the U.S ( EPA 2015; Ramseur 2007). Both the Greenprint plan and the Colorado CAP were preceded by a series of roundtable discussions and public input sessions that p rovided both the formal and informal discussions between advocates and opponents of citywide climate policies. S ince the inception of both plans Mayor Hickenlooper has become the Governor of Colorado and is now leading the charge on the state CAP. The pol icy actors from across Colorado, including those from Denver, will be treated together as subnational policy actors for the following reasons: 1) th e overlap in city and state political actors ; 2) Denver is the largest city in Colorado, the state's capital, and is influential in state

PAGE 35

! 21 level policy developments in climate and energy issues ; and 3) often, as in this case, state and cit y governments cooperate on energy and climate policy Because the D enver and Colorado plan s are consistent with the sets of city and state cli mate action plans in existence ( EPA 2015; Ramseur 2007) a typical case approach to case selection is, according to Gerring (2007, p p 91 93), useful for a study such as this MET HODS: DATA COLLECTION AND ANALYSIS Th is study is an investigation of the climate and energy policy subsystem of Colorado. This study uses the ACF to investigate change and reinforcement of beliefs and advocacy strategies at the individual level using data collected from a 2011 original cross sectional survey. The unit of analysis, as is consistent with similar applications of the ACF in the academic literature cited above, is the individual policy actor active in this subsystem, such as governme nt employees for instance To insure reliability and replicability of this study, the research steps will be explicated below, along with validity issues. Data Collection At the first step, a small number of preliminary open ended i nterviews were conducted with key policy actors to learn about the history of the sub system, refine the instruments and introduce this project to the climate change policy community. The i nterviewees were asked for feedback on a draft version of the survey which help ed to deter mine survey content, especially as it relates to specific policy options and details. These individuals were not a formal advisory council, but aided in the primary goal of insuring the final surveys would be accurate, relevant, and have good face validity in terms of variable measurement

PAGE 36

! 22 The next step in the data collection process was to formalize the target population for the survey. As is indicated and consistent with the literature, policy actors of the climate and energy policy subsystem in Colorado were the target population of this study. Policy actors (sometimes referred to as policy participants or policy elites) are defined as individuals involved in the policy subsystem The individuals were identified as being employed by or involved with government agencies, non governmental organizations, and private companies participating in climate policy making in Colorado Three strategies were used in generating a list of study participants. The first group target ed for inclusion in the study was the individuals who attended the roundtable sessions preceding the creation of state and local government policies and those who attended relevant public i nput sessions during the policy making process. Second, as is consi stent with applications of the ACF, a snowball sample was generated from the lists with input from the preliminary interviewees Third, the major governmental and non governmental organizations in Colorado involved in climate and energy policy were identi fied and the directors, board members, and key staff or officials were also targeted for inclusion These sampling strategies have been used successfully in much of the literature cited above in applying ACF theories to policy subsystems. The completion o f the survey was step two of the data collection process As suggested by Fowler (2002) and mentioned above, the survey was beta tested on the informal preliminary interviewees to confirm face validity and insure clarity and comprehension. Knowing t he expected response rate of the final questionnaire could be approximately 30 60 percent, the range of response rate s of other studies employin g similar techniques (Fowler, 2002) S urveys were e mailed to enough policy actors to attain

PAGE 37

! 23 at least a 30 percent r esponse rate, a goal set to ca pture adequate variation in the sample population In order to increase the response rate, t he survey was designed to take only approximately 20 30 minutes to complete and t imely reminders were sent. In the end, the survey wa s sent to 793 policy actors involved in climate and energy issues in Colorado and 260 individuals returned fully completed surveys (response rate of 33 percent ) The sampling technique that was employed for the survey is explicitly purposive and non probability based as is consistent with the above cited literature. The reason for th e use of this technique is that the views and opinion of active policy actors is explicitly the goal of this study. Data Analysis The quantitative data collected with the surveys were analyzed using STATA 13 software. Descriptive s tatistics and crosstabs were run summarizing policy actor attributes and learning among the sampl e and are presented in Chapter Two Multinominal logistic regression (MNLR) analysis was used to test the effect s extremeness of beliefs, policy activities and sources of learning on policy actor policy learning as belief chan ge or reinforcement in Chapter Three MNLR models were also used to test the effects of extremeness of beliefs, poli cy activities, and sources of learning on policy actor political learning as advocacy strategy change or reinforcement in Ch apter Four The normality, skewedness, and relevant assumptions regarding the tests were examined, as well as the potential for any interactions between independent variables. CONTRIBUTIONS Primarily, this dissertation aims to contribute to the policy process literature. Specifically, t his study attempts to improve existing models and theories of policy actor

PAGE 38

! 24 learning regarding policy beliefs and advocacy strategies. In this way this study hopes to be part of a research effort that advances the theories and measures of learning using the ACF and policy process theories more broadly ( Kraft & Furlong 2015; Birkland, 2011; Freeman, 2006) Building on previous AC F applications this study aims to better understand factors affecting learning. By examining the role of extremeness of belief, policy activities and sources of learning Chapter Three of this study advances research trying to understand the factors that shape different forms of social and instrumental policy learning. In Chapter Four, t he effect of policy actors' extremeness of belief, policy activities and sources of learning on political learning are measured and this w ill also add to the landscape of policy process theorizing. This research will hopefully inform debates regarding potential revisions to established ACF policy belief and policy learning hypotheses. Similarly, as ACF scholars attempt to build better models regarding changes to coalition and individual resources and strategies, the findings here may help improve our understanding of individual political learning. Beyond the ACF, this dissertation also hopes to inform other policy process frameworks. In the s ense that subsystem contextual factors shaping policy actor learning are explored, the findings here may also be useful to scholars using Diffusion of Innovation models of policy process in that pathways to policy learning and political learning are an im portant ca u sal factor in policy diffusion (Berry & Berry, 2014). Similarly scholars using the Institutional Analysis and Development Framework may find the results here useful in their attempts to explain how different sets of actors learn differently dep ending on the context of the policy setting (e.g. rules and norms that may determine frequency of different policy activities) ( Ostrom, Cox & Schlager, 2014).

PAGE 39

! 25 Finally, this dissertation hopes to inform the growing literature examining public policy throug h the lens of the Social Ecological Systems Framework. Understanding the factors that shape individual policy actor learning is important in the creation and implementation of environmental policy as it relates to sustainable management of human used natur al resource systems such as watersheds, fisheries and forests ( Ostrom, Cox & Schlager, 2014; W eible et al., 2010 ). These human used natural resource systems are embedded in complex, social ecological systems (SESs) and contain multiple nested factors across spatial and temporal scales which can be compared to the relationships between organisms, organs, tissues, and individuals cells (Ostrom, 2009). For example, the population living within the jurisdiction of a city h as an environmental footprint including human activities in buildings, transportation patterns, resource consumption, and waste generation. This footprint would be determined, at least in some part, by policy actions at the local level and the aggregated i mpacts of individual behaviors. But this city's SESs would also be embedded within the larger context of trans boundary (cross jurisdictional) infrastructure systems such as electricity generation, as well as broader socio cultural, political, and economi c conditions ( Newell & Cousins, 2014; Ramaswami et al., 2012a 2012b). Changes occur over time in both the biophysical factors as well as the socio cultural factors and t he challenge of our cities and communities being resilient to major changes, as well as the challenge of sustainable development, is dependent on how we learn and respond to these changes ( Schewenius, McPhearson & Elmqvist 2014 ; Anderies, 2015 ; Newell & Cousins, 2014 ; Anderies & Jan ssen, 2013 ) This study will specifically contribute to scholarship and theory building related to policy actor

PAGE 40

! 26 learning and factors that shape learning and specifically in the context of a subnational SES where policy actors are managing energy r esources and climate policy. Sustainability has emerged as both a science and a concept that can be used to organize or reorganize environmental policy discussions. Defined broadly by the United Nation's Brundtland Commission as the ability for a generation to meet its needs w ithout compromising the needs of future generation s, sustainability has become a worthy if not essential goal in our management of complex SESs ( Anderies, 2015 ; Anderies & Janssen, 2013; Dietz et al., 2003) The science and concept of sustainability also provides a lens through which to view environmental problems, policy alternatives and learning (Henry, 2009). But striving for sustainability does not mitigate value clashes in the policy process discussed; it merely provides a new vocabulary and highlig hts the need for diverse viewpoints to engage complex environmental problems in adaptive and collaborative ways. Policy responses to environmental problems such as climate change will require collective action and cooperation across numerous sectors of the econom y and society more broadly. T his will require deliberation among and between various stakeholder groups that may have much to learn from each other in terms of the social dimensions of policy problems such as causal mechanisms as well as political feasibility of policy solutions In other words, the science of sustainability highlights the need for policy actor policy learning and political learning, and better understanding of learning pr ocesses in the policy process. Moving t oward sustainability demands that scientific and technical data be examined and weighed along with the values we hold as a society regarding environmental, economic, and social issues, and that we learn to adapt environment

PAGE 41

! 27 management techniques in ways th at can respond to changes in all of these factors ( Anderies & Janssen, 2013 ; Anderies, 2015 ; Weible et al., 2010; Dietz et al., 2003 ) In this way, learning in the policymaking process is a potential pathway toward collective action, which is in turn a pot ential pathway toward sustainable management of common resources. Thus, if we are to work toward sustainable and adaptive management of SESs and respond to changes in biophysical and socio cultural factors over time, we must understand the influences that shape learning processes ( Anderies & Janssen, 2013 ; Anderies, 2015 ; Gerlak & Heikkila, 2011). Even m ore broadly, by refining models of individual policy learning and political learning, this dissertation ambitiously hopes to further the attempts to better understand learning as a potential indicator of societal benefit of policy process es (Weible, 2014a). Attempts will be immediately undertaken to distribute via publication the findings of this dissertation. Chapter s Three and Four will be revised as part of the goal to create two separate articles for potential publication in venues such as the Policy Studies Journal Review of Policy Research or Policy Sciences Chapter Two may be a candidate for publication in an environmental or climat e change oriented practitioner journal.

PAGE 42

! 28 CHAPTER II S UMMARY ANALYSIS OF CLIMATE AND ENERGY POLICY ACTORS IN COLORADO CHAPTER SUMMARY T he creation of effective and sustainable climate and energy policy at any level requires that policy actors have the ability to learn while adapting to a constantly changing environment. The objective of this chapter is twofold: first, to understand and describe attributes of the policy actors involved in climate and energy policy debates in Colorado including their bel iefs across a number of relevant policy questions and second, to determine the extent of policy learning, as defined as a range between belief change and belief reinforcement, and the extent of political learning, as defined by a change or reinforcement of advocacy strategies, i n the sampled population of policy actors. Colorado is a good case study because, like more than 30 states in the U.S., Colorado launched an initiative to address climate change, which resulted in the creation of the Colorado Clima te Action Plan (CAP) in November 2007. The Colorado CAP is typical in scale, scope, and goals compared to other states. In the spring of 2011, a survey was emailed to 793 policy actors involved in climate and energy issues in Colorado and 260 individuals returned fully completed surveys The results of the survey offer a descriptive portrayal of the diversity in respondents' attributes and range of beliefs regarding the cause, severity, and policy solutions needed to address climate change. Overall, the re sults indicated that these policy actors work in a relatively insular man ner in terms of information sources and policy activities. In terms of policy learning and political learning policy belief reinforcement and advocacy strategy

PAGE 43

! 29 reinforcement were rep o rted with much higher frequency than policy belief change o r change in advocacy strategy. INTRODUCTION The creation of effective and sustainable climate and energy policy at any level requires that policy actors have the ability to learn over time from experience in the policymaking process and be flexible and responsive to changes in the nature of the pol icy problem and politics at play Energy and climate policy actors at all levels of government make decisions based on available scientific evidence, for instance regarding the threats posed by climate change. Policy actors' ability to generate and use rel evant information may vary however. Likewise, policy actors are presumably at least aware of a variety of opinions regarding policy solutions preferred by others in their work, but the extent to which policy actors actually engage in policy activities wit h individuals with contrary belief s likely varies as well. Understanding attributes such as resources, activities, network contacts, and beliefs of policy actors is essential to understand the policy process ( Birkland, 2011; Jenkins Smith et al ., 2014). T he two main objective s guide this chapter to: 1) understand and describe attributes of the policy actors ( individuals in the government, private, nonprofit, and academic sectors) involved in climate and energy policy debates in Colorado including their be liefs across a number of relevant policy questions ; and 2) to determine the extent of policy learning, as defined as belief change and belief reinforcement, and extent of political learning, as defined by a change or reinforcement of advocacy strategies, in the sampled population of policy actors. Despite years of effort

PAGE 44

! 30 in studying learning in public policy, better definitions and measurements of learning is a critical task for scholars o f policy processes (Heikkila & Gerlak 2013). L earning can be though t of as the successive changes or reinforcements in the beliefs or behaviors of individuals regarding the ir perception s of problems and related policies resulting from experience in the policy process ( Jenkins Smith et al., 2014; Birkland 2011 ). For effective decision making, the propensity for learning is linked to policy capacity along with other attributes of policy actors such as policy activities and beliefs. P olicy capacity refer s to the analytical and administrative abilities and skills of the state and local governments as well other non governmental organizations (business and non profits e.g.) to adequately generate and use information to respond to issues in the energy and climate sector (Elgin & Weible 2013). As policy capacity will be me asured as one of the policy actor attributes in this chapter, this analysis builds on previous analyses of policy actors and policy capacity in Colorado (Elgin et al. 2012; Weible et al. 2012 ). In the absence of comprehensive and robust federal climate a ction planning, a patchwork of state and local government polices has been created over the last decade. Like over 30 other U.S. states, Colorado launched an initiative to address climate change, which resulted in the creation of the Colorado Climate Action Plan in November 2007 (EPA 2015) This plan called for a reduction of the state emission of greenhouse gases by 20 percent below 2005 levels by 2020 and a longer term goal of 80 percent below 2005 levels by 2050 ( Colorado Climate Action Plan: A Strategy to Address Global Warming, 2007 ) In terms of scale and scope of policies as well as the shorter term emission reduction targets, the Colorado CAP is typical of state CAPs (EPA 2015;

PAGE 45

! 31 Ramseur 2007). Given this, Colorado could be thought of as a typical case in terms of case selection (Gerring 2007). Colorado represents a state in a transition to what former Governor Ritter referred to as a "New Energy Economy" ( Colorado CAP 2007) The state has a long history of traditional foss il fuel energy production and is now currently experiencing large growth in energy industries, like hydraulic fracturing and unconventional oil and gas development in addition to the less carbon emission intensive renewable energy sectors such as wind an d solar (Lyng 2010) There is reason to believe Colorado policy actors may be especially sensitive to the intersections of energy policy and the threats p osed by climate change, which will be discussed below. The creation and subsequent implementation of this plan involves the participation of policy actors with diverse beliefs. During the participation in this policy process, the policy actors involved will have the potential to learn regarding the social construction of the problem and the relative effec tiveness of different policies (policy learning) as well as the viability and palatability of different policy solutions (political learning). Learning is argued to be a critical component to sustainably managing complex social ecological systems (SESs) ( Anderies, 2015 ; Anderies & Janssen, 2013; Dietz et al., 2003). In that stabilizing and maintaining our climate through sustainable climate and energy policy is essentially the management of an SES, understanding both policy learning and political learning in the relevant policy actors is essential. The focus of this chapter is a ssessing diversity across a host of policy actor attributes, and the extent to which policy learning and political learning are occurring in this policymaking setting during the imp lementation of the Colorado CAP Given that this chapter aims to measure

PAGE 46

! 32 climate and energy policy actor attributes and potential learning this case (Colorado) would seem to be a positive case, a place where the intended measured phenomena will exist (Goer tz 2006). The objective s of this chapter were met using a survey administered in 2011 to 260 policy actors involved in climate and energy issues in Colorado. These policy actors were asked to answer questions related to their particular training and educa tion, organizational affiliation, day to day work, policy related activities, and climate and energy policy related beliefs. The respondents were also asked about policy learning, the extent to which their policy beliefs have either changed or been reinforced, and about political learning, regarding changes or reinforcement of their advocacy strategies. This chapter does not establish a relationship among variables for explanatory empirical testing, but rather provides a discussion and descri ption (relative distribution and patterns) of measured variables in order to better understand the policy actors in the sample group and provides the context of climate and energy issues in Colorado. The Advocacy Coalition Framework (ACF) is used to guide the discussion regarding which attributes of policy actors should be measured. This chapter will set the stage for the two following quantitative analysis chapters focusing on policy learning and political learning respectively THE ADVOCACY COALITION FRAMEWORK The study of learning can be approached from a diverse set of policy process theories, models and frameworks ( Birkland, 2011; Freeman 2006). Some stress the mechanics of diffusion of new ideas in policymaking ( Berry & Berry 2014) while other s examine the way problems and policies are socially constructed (given collective

PAGE 47

! 33 meaning), and how those constructions affect policy desig n and vise versa (Schneider & Ingram 1993 1997). Scholars working with the Institutional Analysis and Development F ramework, building on the work of Elinor Ostrom (2005), are developing theories and models of how institutional design may shape individual and collective learning in policymaking (Heikkila & Gerlak 2013). Lastly, t he Advo cacy Coalition Framework (ACF) ha s become a leading framework to directly examine and exp lain the role of sc ience and technical information, interaction among and between policy actors with similar and different goals, individual belief s and political behaviors and learning ( Jenkins Smith et al ., 2014; Birkland 2011; Freeman 2006). This chapter is guided by the ACF, which was developed by Sabatier and Jenkins Smith after a long history of rese arch on environmental problems and policymaking ( Sabatier 1988; Jenkins Smith et al ., 2014 ). The ACF w as applied to study the policy actors in the climate and energy policy subsystem 4 of Colorado meaning those who were active within Colorado and Colorado cities, for example the City and County of Denver. In applying the ACF in the analys is of a political issue, the focus wa s directed toward the attributes of policy actors, for instance resources, activities, and beliefs, in addi tion to the extent of learning. The reason to focus on individual policy actors comes from the ACF argument that individual policy actor beliefs are the main driver of political behavior and coalesce like minded actors together in to advocacy coalitions (Jenkins Smith et al ., 2014) These advocacy coalitions can be groups of legislators, agency officials, interes t group leaders, !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! !!!!!!!!!!!!!!! 4 A policy subsystem is the larger context of allied and competing groups (interest groups, institutions, and governments) and policy actors involved in the policymaking process in a specific or specialized top ic (Jenkins Smith et al. 2014).

PAGE 48

! 34 and researchers with similar policy beliefs which are normative and empirical beliefs bound by the territorial and topical components of the specific policy issue at hand. For instance, a group of policy actors may share the belief that c limate change will have devastating impacts to communities if left unchecked by government intervention. These groups or the coalition, to varying extents, may share resources and engage in some coordination. The findings summarized in this chapter are not designed to assess the existence of coalitions, but to use the ACF to help assess the diversity of attributes, including but not limited to policy beliefs across the sampled population of policy actors and to examine the extent of learning that has occu rred among these policy actors. Besides policy beliefs, the ACF postulates that other policy actor attributes are important to measure alongside the extent of learning. A policy actor attribute is a quality or characteristic that might vary within the popu lation. For instance the organization affiliation of policy actors, the level of their respective educations, the amount of time spent in the policymaking field, and their respective information sources used in their policy work may be important in the lea rning process ( Sabatier & Jenkins Smith 1999 ; Jenkins Smith et al ., 2014). Likewise, the extent to which policy actors engage in specific policy activities for example c oalition building, coordination and collaborati on within and across coalitions, and participation in multi stakeholder and/or consensus based processes, are critical to assess in terms of understanding this population and the extent of learning that has occurred (Sabatier & Jenkins Smith 1999 ; Jenkins Smith et al ., 2014). This chapter summariz es policy actors' engage ment in specific policy activities together with other policy actor attributes measured in the survey, and the extent of policy and political learning.

PAGE 49

! 35 POLICY ACTOR ATTRIBUTES MEASURED This section will detail the concepts and measures from the survey and provide a short justification of their potential importance in relation to learning but t his chapter will not be testing any specific hypotheses. Examining the diversity of policy actors attributes in terms of: demographics, policy activities, policy capacity, information sources, and policy beliefs will help map the political landscape in this policy subsystem. The operati onal measures for each variable (the specific survey questions used to measure each attribute) are provided in a table in the methodology section below. Demographics of Policy Actors Demographic attributes (qualities or characteristics), for example gender, race/ethnicity, and political orientation were measured to describe the sampled population of policy actors. Other policy actor demographic attributes may have a more intuitive link to learning than gender or race/ethnicity for instance, organizat ional affiliation (Jenkins Smith et al ., 2014). Organization affiliation is the type of organization, or sector, in which the policy actors is predominantly employed. Four major organizational sectors are represented in the sample of policy actors: governm ent employees, individuals from the private sector, individuals from the nonprofit sector, and researchers/academics. There is reason to believe that policy actors' organization or company affiliation may be a factor in shaping learning. For instance, policy actors from government and administrative agencies may posses s more moderate and flexible (or, conversely, less extreme) policy beliefs than actors from non profit environmental groups or private sector oil and gas companies (Jenkins Smith et al. 2 014).

PAGE 50

! 36 Other demographic attributes examined and described in this chapter are the level of f ormal education and time (as measured in years) that policy actors have been involve d in climate and energy policy. These attributes were measured on the survey to help provide an overview of the diversity of whom these policy actors are to help understand the politics on this issue Causal relationships between some of these attributes and learning will be examined in following chapters. Policy Activities Policy ac tivities are defined as any actions or behaviors individual actors engage in as part of their policymaking work 5 The survey asked individuals whether, and how frequently, they participated in a host of policy activities. Research in policymaking processes specifically in the ACF literature, suggest s a few potential relationships between policy actor activities and policy learning (Jenkins Smith et al ., 2014). For instance, individuals that report they seek advice from those with similar beliefs with great er frequency than those that seek advice from those with dissimilar beliefs may suggest the individual has a relatively insular political network, potentially limiting the new ideas and information that might stimulate belief change ( Sabatier & Jenkins Smi th 1999 ) Relatedly, individuals that have participated in more facilitated and consensus based processes may have been exposed to a variety of viewpoints on energy and climate policy issues in a professional setting with established rules a nd norms. Thes e individuals might then report more belief or political behavioral change, as opposed to reinforcement, as their polit ical networks are less insular (Jenkins Smith et al ., 2014). !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! !!!!!!!!!!!!!!! 5 Involvement in the decisions of a government or other authority regarding, for example laws, regulations, programs, and executive functions (Bir kland, 2011).

PAGE 51

! 37 Policy Capacity Policy capacity has been described as the individual and/or organizational ability to acquire and use knowledge in policymaking (Howlett 2009). Therefore, for purposes of this chapter, policy capacity is defined as the analytical and administrative abilities and skills of the policy actors i nvolved in a policy process to adequately generate and use information to respond to issues in the energy and climate sector (Elgin & Weible 2013). Important factors to consider in measuring policy capacity would be: the perceived priority of the issue wi th the respective organization ( reflective of institutional resources and capacity provided by leadership ) the extent to which the government had the knowledge and skillsets needed to respond to policy issues in terms of staff and resources, and the abili ty in engage in long term planning needed in the maintenance of public policy (Elgin & Weible 2013). Individuals were asked to report various measures of their, and their organizations' policy capacity. Policy capacity is examined here as part of the ene rgy and climate policy landscape, and descriptive of the policy actors therein, but the remainder of the broader study will focus on examining learning, a s opposed to policy capacity. Information Sources Policymaking work requires the acquisition and use o f various sources of information. Information sources could be the different authorities, types of documents, and materials used by policy actors in their climate and energy policy related work. Policy actors' information sources are expected to vary in te rms of diversity of information sources and frequency of use of different information sources. Understanding the frequency policy actors utilize different information sources in their work will help

PAGE 52

! 38 understand the role of scientific and technical informati on in this policy subsystem. Policy actors' diversity of information sources, or frequency of use of different information sources may be a factor that impacts learning. Policy Beliefs T he ACF postulates that individual policy actors seek to influence the policy process and that the respective individual policy beliefs are the main driver s of their beha vior (Sabatier & Jenkins Smith 1999). Therefore beliefs are important to understand in an examination of a policy area in terms of some of the politic s that may be at play. Policy beliefs can be defined as a collection of normative and empirical beliefs spanning the substantive policy topic (in this case climate and energy policy) in a geographic area (in this case Colorado) ( Sabatier & Jenkins Smith 1 993 199 9 ). Policy actors in the climate and energy policy subsystem are expected to have a variety of beliefs regarding a host of policy related issues. These policy beliefs are considered to be resistant, but not impo ssible, to change (Sabatier & Jenkins Smith 1999 ) Understanding the shape and scope of diversity of the policy beliefs in the sample of the policy actors is essential therefore in terms of being able to describe this group, and to understand what learning has occurred. The survey particip ants were asked about their relative agreement or disagreement with six policy belief questions regarding major aspects of climate and energy policy and policy preferences. Extreme policy beliefs (e.g. extreme agreement or disagreement) may vary by organi zational affiliation, but also may be a potential attribute shaping learning. The relationships between organizational affiliation and extreme beliefs, and the relationship between extreme beliefs and learning will both be examined in following chapters

PAGE 53

! 39 Policy and Political Learning While no singular conceptual or operational definition of learning across the social sciences exists, learning is often defined as a normative goal and perhaps a prerequisite for collaborative and effective policymaking ( Henry, 2009; Freeman 2006). It is therefore important to think through some different conceptualizations of learning. Bennett and Howlett's (1992) effort to synthesize theories of learning in public affairs parsed three major dimensions of learning: 1) wh o is learning, 2) what is being learned, and 3) what are the results of learning. For example, policy actors may learn about the causal mechanisms of climate change, potentially leading to policy change. I t is important also to differentiate between variou s definitions of what the outcomes or results of learning might be For instance, the outcomes of learning may be conceptualized as broad policy change (Busenberg 2001) while others suggest the results of learning might be individual and/or group belief changes that may or may not lead to policy change ( Sabatier & Jenkins Smith 1999 ; Jenkins Smith et al ., 2014 ). Still others have differentiated between cognitive and behavioral changes as measures of learning ( Gerlak & Heikkila 2011). This final differentiation leads to another often cited typology of learning in public policy that can be used to explore different dimensions of learning. May's (1992) article linking policy learning and failure differentiated two major aspects of learning in public policy. First, policy learning is concerned with two major sub dimensions of learning: 1) social policy learning regarding changes in the social construction of a policy or problem (such as humans as a d r iving cause of climate change or the relative need for the government to intervene to address climate change ) ; and 2) instrumental policy learning regarding changes to the viability or effectiveness of

PAGE 54

! 40 different policy implementation designs to reach existing goals (such as the need for a cap and trade sy stem or a carbon tax to address climate change). The second major dimension of learning May (1992) parses is political learning regarding changes in advocacy strategy for specific policy goals (such as a shift in arguments or tactics). This chapter, as wel l as the larger study therein, will examine both p olicy and political learning. Learning has been explicitly argued to be especially relevant in the policymaking process in the ACF theorizing where it is considered to be an important pathway to policy cha nge (Sabatier 1988). In the context of the policy process, the ACF has defined learning as ongoing adjustments to thought or behavior related to attainment of policy goals from new experiences and/or evidence (Sabatier & Jenkins Smith, 1999, p. 123) From this definition, learning is typically considered to be associated with changes in beliefs or behavior, but might policy actors also reinforce their policy beliefs or advocacy behaviors ? In other words, individual belief or behavioral change is not the only measure of learning. B elief or behavioral reinforcement may occur when the experiences and activities of policy actors serve to galvanize and strengthen their policy beliefs and political activities If inherent value and belief clashes in policym aking lead to intractability, a reinforcement of beliefs and advocacy tactics would seem to only worsen the sclerotic policy process. Conversely, perhaps more change in beliefs and advocacy behaviors during the policymaking process, especially change in a more moderate direction, might lead to more consensus and generate meaning ful policy outcomes If belief and behavioral change increased understanding among stakeholders with differing values, and

PAGE 55

! 41 did not lead to more polarization, perhaps sustainable and adaptive climate policy may be more likely to move forward. While consensus is not the only pathway to policy change, nor is consensus a normative goal, a better understanding the factors that shap e policy ac tor learning is needed. Therefore, delineating between belief change and belief reinforcement, as well as between change and reinforcement of advocacy strategy seems essential in understanding the full picture of what kind of policy learning is taking place in this case. While both belief and behavioral change and reinforcement may be included in a concept of learning in the ACF, this chapter (as well as the broader study ) argues that change and reinforcement are different kinds of learning, diff erent dimensions of the concept of learni ng (Goertz 2006). Specifically examining policy belief and advocacy strategy reinforcement as a dimension of learning in local and state policy actors in the climate and energy policy field is one of the signific ant contributions of the findings described in this study P olicy learning is defined here as both belief change and belief reinforcement similarly political learning is defined as advocacy strategy change and reinforcement. In other words belief and behavioral change and belief and behavioral reinforcement are considered different categories of learning in this study, calling for measuring the relative occurrence of each as well as the absence of either belief or behavioral change and b elief or behavioral reinforcement. Reinforcement can be conceptualized as unchanged but more firmly held belief or advocacy strategy, further buttressed and supported by information or experience. This is different from a change in belief, for instance t oward a more moderate stance, or a change in advocacy strategy, such as a shift in media strategy

PAGE 56

! 42 In a ddition to examining the occurrence learning to better understand and describe the individuals included in this sample, this chapter will include a description of measurement of a range of policy actor attributes along with a discussion of measured variables to provide justification in later analysis However, this chapter will not include an examination of relationships for empirical testi ng. CLIMATE AND ENERGY POLICY IN COLORADO Colorado's history of climate and energy policymaking did not begin with the 2007 Climate Action Plan. In 1994 dedicated employees of The Air Pollution Control Division of the Colorado Department of Public Health and Environment (DPHE) applied and received over $100,000 in Environmental Protection Agency funding to conduct a state greenhouse gas emission inventory (Rabe, 2004). This was done so with an attempt to frame the relatively new policy issue of climate change as linked with long term air pollution control and abatement efforts ongoing in the state (Rabe, 2004). Even before this, the City and County of Denver had demonstrated national leadership by becoming an initial member of the International Coalition of Local Environmental Initiatives (now known as ICLEI) urban greenhouse gas emission reduction program (Bulkeley & Betsill, 2003). A 1991 City ordinance and 1995 Mayoral Proclamation joining the Cities for Climate Protection program, both during the term of then Mayor Wellington Webb, earned Denver a Clean Cities designation, only the second U.S. city to do so (Bulkeley & Betsill, 2003). Mayor Webb, who in his attempts to be viewed as an environmental leader and could be viewed here as a policy entrepreneur also attempted to frame climate change i ssues as part of broader air quality issues in the Ci ty (Bulkeley & Betsill, 2003).

PAGE 57

! 43 Politics soon shifted on this issue however. When an initial state action plan to reduce greenhouse gas emission was drafted in 1997, the plan called for increased regulat ions on the electric utilities sector (Rabe, 2004). This triggered a political backlash from the coal industry, which at the time dominated the energy generation mix in Colorado (Rabe, 2004). The result of this backlash was that Colorado became one of 16 s tates in 1997 and 1998 that passed legislative resolutions expressing: concern over the local economic impact of greenhouse gas mitigation efforts, criticism of the Kyoto Protocol (adopted in December of 1997), and opposition to ratification of the interna tional agreement by the U.S. Senate (Rabe, 2004). While the Colorado legislative resolution attempting to restrict any state actions designed to reduce greenhouse gas emission was never turned into binding legislation, it slowed DPHE efforts to enact modes t greenhouse gas emission reduction plans and demonstrated that the energy industry and other interest groups were capable of framing climate change as an economic issue (Rabe, 2004). Now, Colorado represents a state in a transition to what former Governor Ritter referred to as a "New Energy Economy" ( Colorado Climate Action Plan: A Strategy to Address Global Warming, 2007) Th e state has a history of traditional oil production and is now currently experiencing large growth in multiple energy industries, l ike natural gas development, and less carbon emission intensive r enewable energy sectors, wind and solar energy for instance Many of these recent changes in the energy portfolio of Colorado have been due to state and local policy d r ivers including the Colorado Climate Action Plan (2007), the Clean Air Cool Jobs (2010), and the 2004 establishment of the state's renewable energy standard, which was strengthened by Governor Ritter in 2010

PAGE 58

! 44 opening a new policy landscape and creating new spaces and oppo rtunities for private sector investment and growth (Lyng 2010). There is reason to believe Colorado policy actors may be especially sensitive to the intersections of energy policy and the threats posed by climate change such as: shorter and warmer winters creating a thinner snowpack which may jeopardize the important skiing tourism industry, earlier melting of the snowpack with increased spring runoff leading to significant flash flooding, increased periods of drought and increases in the number of wildfi res, and substantial losses of alpine forests due to pine beetle infestations ( Colorado Climate Action Plan 2007 ) Similar to more than 30 states and 1,000 cities in the U.S., Colorado launched an initiative to address climate change, which resulted in the creation of the aforementioned Colorado Climate Action Plan in November 2007. This plan called for a reduction of the state emission of greenhouse gases b y 20 percent by 2020. Like other CAPs in the U.S., Colorado's climate action plan outlines the package of policies that will be employed to address climate change by making specific policy actions toward reduc ing the GHG emissions This Colorado Climate Ac tion Plan (CAP) is similar in scale and scope to the over 30 other state CAPs in the U.S. ( EPA 2015; Ramseur 2007). METHODOLOGY Research Design Th e findings presented in this chapter are based on a survey conducted in the spring of 2011. The survey was emailed to 793 individuals who were knowledgeable of and actively involved in climate related issues and energy policies in Colorado. Two hundred and sixty (260) respondents filled out the survey completely for a response rate

PAGE 59

! 45 of 33 percent, and the population reflects a broad and diverse set of policy participants. The individuals were selected for participation using a combination of a modified snowball sampling technique, which began wit h current and former staff from the Governor's office to see who was involved in the creation of the state CAP. I nternet research focusing on related policy documents was also conducted to select additional individuals from nongovernment organizations and business es who were involved in climate and energy policy issues. The fourth tactic used to snowball the sample was to include in the invitation to be surveyed a request for additional names of stakeholders involved in climate and energy issues. Operat ional Measures of Policy Actor Attributes T able 2. 1 below provides each of the specific survey questions used to measure each of the policy actor attribu tes examined in this chapter. Table 2. 1 Operational Measures of Attributes Operational Measure s / Survey Question s: Demographics Respondent s were asked to identify their: Gender ( m ale or f emale) Race/ethnicity (American Indian or Alaska Native, Asian, Black or African American, Hispanic or Latino, Native Hawaiian or Other Pacific Islander, and White not of Hisp anic origin). How liberal or conservative they consider themselves to be on fiscal policy (on a scale of: very liberal, liberal, moderate, conser vative, and very conservative) How liberal or conservative they consider themselves to be on s ocial policy (on a scale of: very liberal, liberal, moderate, conser vative, and very conservative) What best describes their organization (academic/research, business/private sector, gov ernment, media, and non profit) Highest level of education attained ( n ot a high school graduate, h igh school graduate, s ome college, b achelor's degree, m aster's or professional degree, and Ph.D./M D /J D ) Years they have been involved in climate related issues and/or energy policy (on a scale of: <1, 1 5, 6 9, 10 14, 15 20, and >20 ) Operational Measure s / Survey Question s: Policy Activities Respondents were asked to report how often they used the following tools and techniques as part of their work in the past year (on a scale of: daily, weekly, monthly, yearly, and never) : Collaboration with those who share their views on energy & climate goals Collaboration with those who share their v iews on energy & climate goals Facilitation/consensus building (e.g. focus groups, roundtables)

PAGE 60

! 46 Table 2.1: Operational Measures of Attributes (cont.) Respondents were asked (yes or no) in the past year, which of the following activities did they participate in: Participated in coalition building (e.g., networking, information sharing) Negotiated in a multi stakeholder consensus base d process Operational Measure s / Survey Question s: Policy Capacity Respondents were asked to report how often they used the following tools and techniques as part of their work in the past year (on a scale of: daily, weekly, monthly, yearly, and never) : Community level impact analysis (e.g. neighborhoo d surveys) Political feasibility analysis (e.g. SWOT analysis, polling data) Risk assessment Modeling (e.g. climate change scen arios, energy futures analysis) Environmental impact analysis Economic and financial analysis (e.g. cost benefit and econo mic impact analysis) Informal tools and techniques (e.g. brainstorming, problem mapping) Respondents were asked to respond (on a scale of: v ery low capacity, l ow capacity, m edium capacity, h igh capacity, and v ery high capacity) to the following questio n: Compared to similar organization s, does your organization have adequate knowledge, skills, and people to respond to climate relate d issues and energy policies? Respondents were asked to respond (on a scale of : m uch lower, l ower, a bout the same, h igher, and m uch hig her) to the following question: Compared with other issues that your organization responds to, how much of a priority are climate relate d issues and energy policies? Respondents were asked ( y es or n o) in which of the following areas have they received formal training: A pplied research Modeling Policy analysis Policy evaluation Statistical methods Trends analysis and/or forecasting Respondents were asked to indicate their level of agreement (on a scale of: strongly disagree, disagree, neither agree nor disagree, agree, and strongly agree ) with the following statement: I have timely access to academic literature, peer reviewed publications and professional research relevant to climate related issues and energy policy work. Respondents were asked ( y es or n o) in the past year, which of the following acti vities did they participate in: Appraised policy options Conducted research on climate related issues and/or energy policy Consulted with the public Ev aluated policy processes, results and outcomes Implemented or delivered policies or programs on climate related issues and/or energy policy Informed e lected and appointed officials

PAGE 61

! 47 Table 2.1: Operational Measures of Attributes (cont.) Respondents were asked, with respect to climate and energy issues, to indicate their relative agreement (on a scale of: strongly agree, agree, neutral, disagree, and strongly disagree) with the following statements: I regularly engage in tasks which demand immediate attention (e.g., "firefighting") I regularly engage in tasks which relate to long term planning (e.g., more than a year) Urgent day to day issues seem to take precedence over thinking "long term" I am provided enough time and resources to undertake tasks and planning that are engaging for more than a year I am increasingly consulting with the public as I do my policy related work Policy decisions seem to increasingly be those that are most politically acceptable There seems to be less governmental capacity to analyze policy options than there used to be My policy related related work increasingly involves networks of people across regions, or levels of government Policy problems increasingly require strong technical expertise Much of th e existing skills and knowledge about climate and energy issues lies outside the formal structure of government Those who have more authority in decision making usually have less specialized technical expertise Operational Measure s / Survey Question s: Inf ormation Sources Respondents we re asked to indicate how often they used the following information sources in their policy work ( on a scale of: daily, weekl y, monthly, yearly, and never): R eports produced /created by their organi zation A dvice from individuals the y agree with A dvice from i ndividuals they disagree with Reports from non profits Personal experience Budgets and cost data Academic research Newspapers and magazines R eports of other government agencies R eports from consultants Industry reports Online social networks (Facebook, Twitter) Operational Measures / Survey Question s: Policy Beliefs Respondents were asked to indicate their rel ative agreement (on a scale of s trongly agree, s omewhat agree, n either agree nor disagree, s omewhat disagree, and s trongly disagree) with the following statements: The severity of predi cted impacts on society from climate change are vastly overstated Human behavior is the pr incipal cause of climate change Decisions about energy and its effect on climate are best left to the economic ma rket, and not to the government An energy and/or carbon tax is re quired to combat climate change A cap and trade system of permits for the emission of greenhouse gasses is required to combat climate change Government policies to promote renewable energy generation are required to combat climate change

PAGE 62

! 48 Table 2.1: Operational Measures of Attributes (cont.) Operational Measure s / Survey Question s: Policy Learning For each of the six policy belief questions above, respondents were asked to what extent their views had changed or been reinforced: Only been reinforced Mostly reinforced with little changes Balance of reinforcement and changes Mostly changed with little reinforcement Only been changed My views have neither changed nor been reinforced Operational Measure s / Survey Question s: Political Learning Respondents were asked to what extent their strategies had been changed or reinforced regarding the way they advocate for climate related issues and/or e nergy policy : Only been reinforced Mostly reinforced with little changes Balance of reinforcement and changes Mostly ch anged with little reinforcement Only be en changed Neither changed nor been reinforced RESULTS Objective 1: Policy Actor Attributes The results pres ented below are broken down by policy actor attribute, policy activities, aspect s of policy capacity policy be liefs, and advocacy coalitions. Demographics The majority of respondents in the sample (54 percent) were female. In terms of race the sample was overwhelmingly white (not of Hispanic origin), with 90 percent of the sample self identifying in that way. Nearly 50 percent of respondents described themselves as moderate on fiscal policy, with a relatively even distribution indicating they were more liberal or more conservative on fiscal policy. O ver 70 percent of respondents reported themselves to be liberal to very liberal on social policy.

PAGE 63

! 49 Figure 2.1 Survey Respondents by Political Orientation Organizational Affiliation Respondents were asked to identify their organization with the following question: "Which of the following best describes your organization?" As seen in F igure 2 .2 below: a third of the respondents were from the business/private sector nearly a third were government employees and the other third worked for non profit organizations and/or identified as working in the fields of academic/research. T wo respondents (<1 percent ) were representatives of the media. "! #"! $"! %"! &"! '""! '#"! '$"! ()*+!,-.)*/0! ,-.)*/0! 123)*/4)! 5267)*8/4-8)! ()*+! 5267)*8/4-8)! !"#$%&'()'*+,-.-,"/01' 2"&.%3'4%15(+,%+61'$3'7(0-6-8/0' 9&-%+6/6-(+'' 9-7:/0!;20-:+! <2:-/0!;20-:+!

PAGE 64

! 50 Figure 2.2 Organizational Affiliation of Survey Respondents Government officials came from city, state, and federal level agencies. Researchers came from colleges and universities, private consulting firms, and government research organizations. Non profit organizations consist largely of environmental groups active in energy and climate related issues in Colorado. Given the low response rate from media re presentatives, they were not included in the analysis below. Formal Education In terms of formal education, 53 of 87 (61 percent) of the respondents from the business sector reported having obtained at least a Masters Professional Degree, post secondary pr ofessional degrees, or higher Sixty five of 79 (82 percent) and 44 of 54 (81 percent) reported the same in the government and non profit sectors respectively =&! '>?! @A! ="?! >$! #'?! &@! ==?! #! '?! 9&:/+-;/6-(+'<)=-0-/6-(+'()'2"&.%3' 4%15(+,%+61>'' ?@A'7(0-83'<86(&1' B)7)/*:CDE:/3)F-:! G28)*6F)64!! H26IJ*2K-4! LM7-6)77!! 1)3-/!

PAGE 65

! 51 I ndividuals who identified as researchers/academics had the highest percentage having earned a Ph.D., M.D., or J.D. at 5 1 percent Figure 2.3 Respondents by Formal Education Years Involved Across all sectors, more than half of survey respondents indicated they had been involved in climate and energy issues for fewer than 10 years, with the most common response 1 5 years In comparison to the other sectors, individuals working for government agencies reported a shorter time of involvement in climate and energy issues, with over 50 percent indicating that they had been involved five years or less. "?! '"?! #"?! ="?! $"?! >"?! %"?! @"?! &"?! A"?! '""?! 7%&8%+6'()'*+,-.-,"/01' 4%15(+,%+61'$3'B(&#/0'C,"8/6-(+' ;CNONP!1O!2*!QO!! 1/74)*R7!2*!;*2S)77-26/0! O)T*))! L/:C)02*R7!O)T*))! <2F)!5200)T)!! U-TC!<:C220!G*/3M/4)!!

PAGE 66

! 52 Figure 2.4 Respondents by Years Participating in Climate and Energy Issues Policy Activities As can be seen in Figure 2. 5 below, r espondents reported they more frequently collaborated with individuals with similar views on climate and energy goals tha n they collaborated with individuals with di ssimilar views on these issues. "?! #"?! $"?! %"?! &"?! '""?! 7%&8%+6'()'*+,-.-,"/01' 4%15(+,%+61'$3'D%/&1'7/&6-8-5/6-+:'-+' E0-#/6%'/+,'C+%&:3'*11"%1' V!#"!W)/*7! '>I#"!W)/*7! '"I'$!W)/*7! %IA!W)/*7! 'I>!W)/*7! X!'!W)/*!

PAGE 67

! 53 Figure 2.5 Respondents by Frequency of Engagement in Policy Activities T he vast majority of respondents reported collaboration with "those who share my views on climate and energy goals" on a much more frequent basis (the most common answers being daily or weekly ) than they report ed collaboration with those "those who do not s hare my views " (the most common answer being monthly ). Participating in facilitation/consensus building was overall less common. A total of 159 of 260 respondents (61 percent) reported they engaged in facilitatio n/consensus building ( focus groups and rou ndtables e.g. ) only monthly or yearly, and 49 respondents (19 percent) reported they never engage in this polic y activity. "! '"! #"! ="! $"! >"! %"! @"! &"! A"! '""! O/-0+!! Y))Z0+! 1264C0+! W)/*0+!! H)8)*! !"#$%&'()'4%15(+,%+61' B&%F"%+83' B&%F"%+83'()'C+/:%#%+6'-+'7(0-83' <86-.-6-%1' 5200/.2*/4)![-4C!\C27)!YC2!
PAGE 68

! 54 The results above comport with additional policy activities related results regarding two yes/no policy activity questions First, 63 percent of policy actors answered, Y es they had participated in "coalition building" ( networking and information sharing e.g. ) in the past year. Second, just over half (56 percent ) indicated Y es they had "negotiated in multi stakeholder consensus base d processes These results together would seem to indicate that policy actors are more likely to engage in policy activities within coalitions of individuals with similar views as opposed to between coalitions of actors with dissimilar views Policy Capacity The analysis of questions measuring policy capacity indicated that t he majority of individuals reported similar levels of experience and formal training in a variety of analytical techniques and use of tools and techniques as part of their work in the previous year It is worth noting that when asked directly about policy capacity, g overnment agency employees reported the lowest levels of organizational policy capacity.

PAGE 69

! 55 Figure 2.6 Extent of Policy Capacity in Respondents' Organization Overall though, policy actors who responded to our survey reported that their organizations had high policy capacity and were able to address climate change and energy issues based on their resources. The relative priority of climate and energy related issues within an organization may indicate the amount of human and technical resources (capacity) allocated by leadership to address those issues. Across sectors, respondents also indicated that climate a nd energy issues were a higher or much higher organizational priority compared to other issues which is perhaps not surprising given that participation in these policy issues was the reason these individual s were selected for the survey. "?! #"?! $"?! %"?! &"?! '""?! 7%&8%+6'()'*+,-.-,"/01' GE(#5/&%,'6('1-#-0/&'(&:/+-;/6-(+1H',(%1' 3("&'(&:/+-;/6-(+'I/.%'/,%F"/6%'J+(K0%,:%H' 1J-001H'/+,'5%(50%'6('&%15(+,'6('80-#/6%L &%0/6%,'-11"%1'/+,'%+%&:3'5(0-8-%1MG'' ()*+!U-TC!! U-TC!! 1)3-MF! ,2[! ()*+!,2[!!

PAGE 70

! 56 Figure 2.7 Pr iority of Climate and Energy Related Issues In terms of the amount of research conducted and accessed, academic and professional researchers reported the highest abilities to conduct research in terms of formal training, while government employees reported the second hig hest ratios of formal training. Table 2.2 Percentage of Respondents That Received Training in Applied Research Percentage of Respondents That Received Training in Applied Research Research / Academic 66% Business 30% Government 37% Non profit 18% Individuals from all sectors reported having timely access to academic and professional research on climate and energy policy, but government employees reported so less frequently than the academic/research of the business sectors. "?! '"?! #"?! ="?! $"?! >"?! %"?! @"?! &"?! A"?! '""?! 7%&8%+6'()'*+,-.-,"/01' 7&-(&-63'()'E0-#/6%'/+,'C+%&:3L4%0/6%,'*11"%1' E(#5/&%,'6('96I%&'*11"%1'-+'9&:/+-;/6-(+' 1M:C!U-TC)*! U-TC)*!! E.2M4!4C)!
PAGE 71

! 57 Figure 2.8 Respondents Reported Access to Professional Research Regarding t he ability to measure the opinions and preferences of the public, interest groups, and major policy players (an important aspect of policy capacity discussed above) c ompared to the other sectors, government agency employees reported the highest frequency of use of community level impact analyses such as neighborhood surveys with the most common response indicating yearly In terms of another important component of policy capacity, t he ability to communicate to stakeholders the survey results indicate t he majority of government employees (51 percent) reported that their policy work increasingly involved consulting with the public, and 71 percent reported the y participated in coalition building (e.g. networking or information sharing) Lastly, summarizing the final set of policy capacity related results of Yes/No questions measuring the ability to articulate medium and long term priorities: a "?! #"?! $"?! %"?! &"?! '""?! 7%&8%+6'()'*+,-.-,"/01' G*'N/.%'O-#%03'<88%11'6('7&()%11-(+/0' 4%1%/&8IG'$3'2%86(&''' <4*26T0+!ET*))! ET*))! H)-4C)*!ET*))!62*!O-7/T*))! O-7/T*))! <4*26T0+!O-7/T*))!

PAGE 72

! 58 majority of government agency e mployees (61 percent) indicated they regularly engaged in long term tasks and 48 percent also indicated they were not provided with the necessary resources and time to engage in long term planning. Additionally, 73 percent of government employees indicated short term issues took precedence over long term thinking. Information Sources The survey respondents were asked how frequently they used 12 sources of information in their climate and energy related policy work. Table 2. 3 below presents a summary of resp onses organized by descending values of daily usage helping to identify what information sources were used most frequently. Two noteworthy items : more than 50 percent of respondents indicated they used "personal experience" daily in their policy work, and more than 50 percent of responde nts indicated they never used "o nline social networks" in their policy work. Also standing out: besides "o nline social networks respondents indicated they used each source of information at least yearly 90 percent of the time. In other words, besides "o nline social networks only 10 percent or less of the time did respondents on average indicate they never used any individual information source. The summary table below thus supports the f ol lowing: a large percentage of re spondents use d the vast majority of given information sources at least to some extent. The summary total does not include a sum column or row due to the fact that sometimes one or two of the 260 respondents skipped (did not answer) one of the information sources, but the total for each source ranged 256 260.

PAGE 73

! 59 Table 2.3 Respondent Information Use Summary Other trends that include "advice from people you agree with" and "reports created by your own organization" we re reported by respondents as used more frequently in their policy work than for instance "advice from people you disagree with or reports from the variou s outside organization s queried. This, combined with the prime use of "personal experience would seem to indicate a relatively insular set of information sources being utilized by the policy actors in this subsystem Policy Beliefs T he policy actors wer e asked to demonstrate their policy beliefs about climate change and energy policy related issues by indicating their relative agreement or disagreem ent with six policy statements. The results are shown in Figure 2.9 below The majority of respondents stro ngly agreed with the statement about the need for policies to promote renewable energy, with the chart of policy beliefs below organized by descending levels of strong agreement with the six statements. Daily Weekly Monthly Yearly Never Personal experience 57% 18% 16% 5% 4% Newspapers and news magazines 42% 26% 22% 1 6% 1 4% Advice from people you agree with 31% 34% 29% 1 4% 1 2% Reports created by your own organization 18% 24% 31% 17% 10% Budget and cost data 18% 26% 31% 15% 10% Advice from people you disagree with 17% 22% 45% 11% 1 5% Academic research 16% 23% 42% 14% 1 5% Reports from industry 13% 27% 37% 19% 1 4% Online social networks 11% 15% 12% 1 7% 54% Reports from other governments 1 9% 26% 43% 15% 1 7% Reports from consultants 1 8% 22% 41% 20% 1 9% Reports from non profits 1 6% 26% 40% 19% 1 9%

PAGE 74

! 60 To summarize, the majority of participants reported they strongly agree or agree that : 1) the impacts of climate change are not overstated ; 2) human behavior is the principal cause of climate change ; 3) decisions about energy a nd its effect on climate are best left to the government, as opposed to the market ; and 4) government polices to promote renewable energy generation are required to combat climate change. Compared to these four policy beliefs, t here was greater diversity of opinions on the need for a carbon tax or a cap and trade system to com bat climate change, but the majority of respondents strongly or somewhat agreed with the need for the former, and respondents were more neutral (more reported neither agree nor disagree ) on the need for the latter.

PAGE 75

! 61 Figure 2.9 Policy Beliefs Advocacy Coalitions Elgin and Weible (2013) conducted a different but related analysis of this same sample of Colorado climate and energy policy actors. While extensive mapping and exploration of the advocacy coalition in this subsystem is not an objective of the analysis presented in this chapter, some discussion of this is warranted. Analyzing the results of the six policy belief questions using cluster analysis and silhouette means Elgin and Weible (2013) found that the actors fell into two advocacy coalitions based on belief systems. A larger proclim a te coalition of 205/260 (79 percent) policy actors believed "?! '"?! #"?! ="?! $"?! >"?! %"?! @"?! &"?! A"?! '""?! G28)*6F)64! ;20-:-)7!42! ;*2F24)! B)6)[/.0)! ]6)*T+! G)6)*/4-26!/*)! B)^M-*)3!42! 52F./4!50-F/4)! 5C/6T)! UMF/6!L)C/8-2*! -7!4C)!;*-6:-J/0! 5/M7)!2S!50-F/4)! 5C/6T)!! E6!]6)*T+!/63D 2*!5/*.26!\/_!-7! B)^M-*)3!42! 52F./4!50-F/4)! 5C/6T)!! E!5/J!`!\*/3)! <+74)F!a7! B)^M-*)3!42! 52F./4!50-F/4)! 5C/6T)!! \C)!;*)3-:4)3! aFJ/:47!S*2F! 50-F/4)!5C/6T)! /*)!b8)*74/4)3!! O):-7-267!/.2M4! ]6)*T+!`!-47! ]SS):4!26!50-F/4)! /*)!L)74!,)S4!42! 4C)!1/*Z)4P!/63! H24!42!4C)! G28)*6F)64! 7%&8%+6'()'*+,-.-,"/01' 7(0-83'P%0-%)1''' <4*26T0+!ET*))!! <2F)[C/4!ET*))!! H)-4C)*!ET*))!H2*!O-7/T*))!! <2F)[C/4!O-7/T*))!! <4*26T0+!O-7/T*))!!

PAGE 76

! 62 that climate change impacts were serious and should be addressed by government intervention, and were supportive of policies aimed at doing so (Elgin & Weible 2013). The anticlimate coalition consisted of 55/260 (21 percent) policy actors with fundamentally opposing policy beliefs (Elgin & Weible, 2013). Besides vastly different policy beliefs across coalitions, two other significant differences related to coalition membership were observed. First, the proclimate coalition had a majority of its members from the government sector, while the majority of anti climate coalition members were private sector employees. Second, the proclimate coalition members identified themselves as significantly more socially and fiscally liberal than the anticlimate members, who identified themselves as more conservative sociall y and f iscally (Elgin & Weible, 2013). Despite opposing policy beliefs and these differences in attributes among policy actors across coalitions, the coalitions themselves were found to be similar in many ways (Elgin & Weible, 2013). The coalitions had sim ilar levels of individual education, experience and training. Policy capacity levels in terms of knowledge, skill, resources, and organizational priority of climate and energy issues were similarly high in both coalitions. Comparatively, very similar poli cy activities and advocacy strategies were used by the two coalition members. Elgin a nd Weible (2013) concluded that, given the proclimate coalition was proportionately larger, and that Colorado supports a Climate Action Plan, there is some reason to belie ve this coalition was in a stronger position and had some policy goal attainment success That said, the researchers make clear that any claims regarding relative influence and absolute success of the two advocacy coalitions in the subsystem would need ad ditional e vidence (Elgin & Weible, 2013).

PAGE 77

! 63 Objective 2: Learning Results presented below provide details regarding policy actors' policy learning and politic al learning in this subsystem. Policy Learning Policy actors were asked to demonstrate their policy learning, defined here as their responses to a scale between reinforcement of beliefs and change in beliefs but also including neither change nor reinforcement of beliefs As can be seen in Figure 2.10 below, in terms of policy learning, policy actors tended to report belief reinforcement with much greater frequency than belief change but neither change nor reinforcement was more common than belief change.

PAGE 78

! 64 Figure 2.10 Policy Learning Across the Six Policy Belief Questions Political Learning Respondents were asked to demonstrate their poli tical learning, defined here as their responses to a scale between reinforcement of advocacy strategies and change in advocacy strategies, but neither change nor reinforcement was included. As can be seen in Table 2.4 below, like with policy learning, in terms of political l earning, respondents "! #"! $"! %"! &"! '""! '#"! 1+!(-)[7!U/8)! b60+!L))6! B)-6S2*:)3!! 1+!(-)[7!U/8)! L))6!12740+! B)-6S2*:)3!Y-4C! ,-440)!5C/6T)7!! 1+!(-)[7!U/8)! U/3!/!L/0/6:)!2S! B)-6S2*:)F)64!`! 5C/6T)7!! 1+!(-)[7!C/8)! 12740+!5C/6T)3! Y-4C!,-440)! B)-6S2*:)F)64!! 1+!(-)[7!U/8)! 5C/6T)3!b60+!! 1+!(-)[7!U/8)! H)-4C)*!5C/6T)3! H2*!L))6! B)-6S2*:)3!! !"#$%&'()'*+,-.-,"/01' 7(0-83'Q%/&+-+:'<8&(11'6I%'2-R'7(0-83' P%0-%)'S"%16-(+1' \C)!;*)3-:4)3!aFJ/:47!S*2F!50-F/4)!5C/6T)!/*)!b8)*74/4)3!! UMF/6!L)C/8-2*!-7!4C)!;*-6:-J/0!5/M7)!2S!50-F/4)!5C/6T)!! O):-7-267!/.2M4!]6)*T+!`!-47!]SS):4!26!50-F/4)!/*)!L)74!,)S4!42!4C)!1/*Z)4c! E6!]6)*T+!/63D2*!5/*.26!\/_!-7!B)^M-*)3!42!52F./4!50-F/4)!5C/6T)!! E!5/J!`!\*/3)!<+74)F!a7!B)^M-*)3!42!52F./4!50-F/4)!5C/6T)!! G28)*6F)64!;20-:-)7!42!;*2F24)!B)6)[/.0)!]6)*T+!G)6)*/4-26!/*)!B)^M-*)3c!

PAGE 79

! 65 tended to report reinforcement of advocacy strategy with muc h greater frequenc y than change in advocacy strategy, though neither chang e nor reinforcement was common. Table 2.4 Political Learning in Survey Respondents 6 Category of Political Learning Regarding Advocacy Strategies Number of Respondents Only Been Reinforced 59 Mostly Reinforced with Little Changes 59 Balance of Reinforcement and Changes 76 Mostly Changes with Little Reinforcement 10 Only Been Changed 1 4 Neither Changed nor Been Reinforced 43 CONCLUSIONS The concept of learning is central to many approaches to studying public policy and policy processes and particularly so within the ACF (Freeman 2006; Birkland 2011 ; Heikkila & Gerlak 2013). Recent meta stu dies of learning (see Murro & Jerry 2008; Reed et al ., 2010; Crona & Parker 2012) agree that despite the lack of a single definition or measure of learning across the social sciences, learning is often cited as a goal and perhaps a prerequisite for effective governance and sustainability It is therefore important to measure the extent of policy actor lea rning and the various forms that learning may take when examining policy subsystems and surveying policy actor attributes in hope of better u nderstanding policy processes. The first major objective of this chapter was to understand and describe attributes of the policy actors involved in climate and energy policy debates in Colorado, including their beliefs across a number of relevant policy questions !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! !!!!!!!!!!!!!!! 6 Note: the total response count for the question measuring political learning was 251, slightly below the 260 total for most of the previous questions examined. This may be in part due to the location of the question toward the end of the survey.

PAGE 80

! 66 The results indicate that this sample of policy actors varies across many attributes including organizational affiliation and a number of policy activities, as well as across the policy beliefs examined. Some of the findings support claims made by the ACF regarding policy actors working within in a policy subsystem detailed below. One third of the sampled population were government employees, another third were from the private sector, and individuals from the non profit sector and academic/research field made up the final third. This organizational diversity of policy actors is consistent with the ACF argument and findings of previous ACF applications that policy subsystems involve more than just government employees in policymaking (Jenkins Smith et al ., 2014). The results of the survey show that the policy actors surveyed were essentially evenly s plit gender wise, predominantly white (not of Hispa nic origin), and well educated. The vast majority of respondents reported more frequent collaboration with policy actors with similar views than collaboration with policy actors with dissimilar views Indi viduals reported specific partici pation in facilitated consensus building e ven le ss frequently than collaboration Engagement in facilitated consen sus building ( focus groups and roundtables e.g. ) was predominantly done on only a monthly or yearly basis, and about a fifth of the respondents reported they never engage in this policy activity at all These results together would seem to indicate that policy actors are more likely to engage in policy activiti es within coalit ions of individuals with similar views as opposed to between coalitions. This may support the findings of a previous ACF application that coalitions of policy actors working together to advance policy goals may be shaped by the nature of perceived opponen t s and allies (Henry, Lubell & McCoy 2011). The

PAGE 81

! 67 frequency of information sources used by the respondents indicates diverse, but also relatively insular and fragmented p olicy information utilization. In terms of policy capacity, results indicate that polic y actors working in this climate and energy policy system believe government has a mixed level of policy capacity in this area. That said, there was consensus that the policy capacity in the public sector on these issues needs to be increased. Given the ta rget population and purposive sampling process our respondents were very likely knowledgeable individuals on this matter In terms of policy beliefs, the results indicate a range of views, b ut broadly speaking, the majority of respondents agreed the effec ts of climate change are serious, that humans are causing climate change, that the government should be involved in climate and energy policy and specifically in the need for policies promoting renewable energy generation. Views on the need for a carbon/energy tax or a cap and trade system were more mixed. This seems consistent with established ACF hypotheses that there is less consensus on specific policy pref erences than on beliefs central to the perception of the problem ; however a recent review of ACF applications found mixed support for these hypotheses (Jenkins Smith et al ., 2014). The second major objective of this study was to determine the extent of po licy learning, as defined as a range between belief change and belief reinforcement, and extent of political learning, defined here as a range between a change and reinforcement of advocacy strategy over time in the sampled population of policy actors In general, across all six major policy beliefs, reports of policy b elief and advocacy strategy reinforcement were much more common than reports of policy belief or advocacy strategy change.

PAGE 82

! 68 The greater incidence of belief and behavioral reinforcement may be of concern as it potentially signals that the individuals involved in policy debates in this context are becoming more intransigent, potentially resulting in their reduced ability to promulgate effective and adaptive climate change in the near future ( Meijerink 2005; Litfin 2000) That said, a compromise of beliefs is not the only pathway for policy change. Relatedly, t he fact that most policy actors report ed relatively insular networks of advice and information sources, as well as relatively less c ollaboration with individuals with dissimilar views or participation in facilitated multi stakeholder consensus building, may indicate a paucity of dialogue between policy actors with different beliefs on the relevant problems and potential policies. T his suggests that the sub system may have trouble building the policy learning and political learning over time needed to create sustainable climate and energy policies (Jenkins Smith et al ., 2014) When this is considered along with the greater prevalence of b elief reinforcement as opposed to belief change among policy actors, this becomes even more troubling. Again, the researcher is not suggesting consensus is a de facto superior policy process outcome but normative arguments regarding a link between learnin g and sustainable policymaking are made elsewhere (Henry, 2009). If a greater effort can be made to increase attendance and meaningful engagement in roundtables and other facilitated consensus building activities, perhaps these trends can be mitig ated or r eversed. It is important to acknowledge the limitations of th ese data when attempting to generalize these findings in the Colorado climate and energy policy system to other policy arenas or geographic areas. For instance, this analysis considered all government levels (state and city for instance) and agencies together Relatedly, given that Denver is

PAGE 83

! 69 the capital city (higher population and greater connectivity to state level government issues) the total number of local government actors is skewed toward the City and County of Denver employees, and less so on small com munities and other cities in Colorado. Further research is required to unpack the difference between different sub populations of government employees, and a similar argument could be made for the business sector, academic/researchers, and non profit emplo yees. Moreover like all self reported opinion based research, the results represent reports potentially limited or biased by our sampling design or the answers provided This chapter presents findings indicat ing this sample of policy actors displays diver sity across many attributes including organizational affiliation, policy activities, and among the six policy beliefs measured. Some of these findings bolster assumptions and arguments made by the ACF and corroborate some previous applications regarding po licy actors, beliefs and various forms of learning ( Jenkins Smith et al., 2014; Henry, Lubell & McCoy, 2011; Meijerink 2005; Litfin, 2000). Additional analysis is required to explore potential relationships between policy actor attributes and possible ef fects on policy actor learning.

PAGE 84

! 70 CHAPTER III F ACTORS SHAPING POLICY LEARNING : A STUDY OF POLICY ACTORS IN SUBNATIONAL CLIMATE AND ENERGY ISSUES CHAPTER SUMMARY T he creation of effective and sustainable climate and energy policy at any level requires that policy actors have the ability to learn Information and experiences are interpreted in ways that may change or reinforce the beliefs of the individuals and groups engaged in the policy process. This change over time has been defined as policy learning a nd the concept of learning has long played a central role in the theories and frameworks used to understand policy processes. Findings described in t his chapter a im to contribute to the theoretical and methodological understanding of individual learning in the policy process by explicitly examining belief change and belief reinforcement as products of policy learning, measuring both, as well as measuring the absence of either. Analysis described in this chapter examine d several factors associated with policy learning including: policy actors extremeness of beliefs, the extent to which policy actors engage in policy activities such as collaboration and advice seeking within and between belief coalitions and facilitated consensus building processes Similar to many U.S. states, Colorado is experiencing changes in energy production patterns regarding oil and gas development and renewable energy generation. Like more tha n 30 states these changes are happening in the context of an initiative to address climate change, the Colorado Climate Action Plan launched in November 2007. The objective of this chapter is to use the lens of the Advocacy Coalition Framework (ACF) to he lp examine some of the factors

PAGE 85

! 71 that promote and shape policy learning in policy actors involved in climate and energy policy debates in Colorado regarding their beliefs across a number of relevant policy questions. In the spring of 2011, a survey administered to climate and energy policy actors in Colorado measured policy learning and the factors that may shape policy learning in these policy actors. The results indicate that extreme beliefs are associated with belief reinforcement, relative to policy actors with more moderate beliefs, and that collaboration with ind ividuals with differing policy views is associated with belief change. INTRODUCTION In democratic societies, the policymaking process can involve the clash of competing values leading to prolonged political intransigence. This in true across many policy ar eas, but can be especially true in e nvironmental policy processes where policy alternatives are explicitly value laden In these cases a diversity of policy beliefs held by individuals may exist across a variety of factors such as: the role of ecosystem protection, economic development, and soci al equity issues. Acquired i nformation and experience are interpreted in ways that may change or reinforce the beliefs of the individual actors engaged in the policy process. This change or reinforcement over time has been d efined as a learning and the concept of l earning has long played a central role in the theories and frameworks used to understand policy processes (Heclo, 1974; Be nnett & Howlett, 1992 ; May, 1992; S abatier & Jenkins Smith, 1993; Gerlak & Heikk ila, 2011 ) In The Science of Muddling Through Charles Lind b lom (1959) an important early contributor to the study of public policy, offered the idea of policymaking as a process of incrementalism a sequence of successive limited comparisons that allow decision makers to engage in learning from previous policy desi gn attempts and failures

PAGE 86

! 72 These successive comparisons occur in the policy process, defined by Sabatier (2007, p. 3) as the process by which " problems are conceptualized and brought to the government for solution; governmental institutions formulate alternatives and select policy solutions; and those solutions get implemented, evaluat ed and revised In Sabatier's terms, policy actor learning occurs as different policy ideas are brought forward and compete in the policy process. The study of policy processes can now be approached from a diverse set of theories and frameworks (Cairney & Heikkila, 2014), and the concept of learning remains an important component to many of these theories a nd frameworks (Heikkila & Gerlak, 2013; Freeman, 2006). But there is still much to examine about the specific approaches to individual learning in policy processes. For instance, different factors may shape policy learning in different ways. In other words there may be different products (or results) of policy learning, such as policy belief change or belief reinforcement based on the attributes and experiences (the learning process) of individual actors involved in the policy process. In terms of policy learning as belief change and belief reinforcement, some of the factors that have been recognized as shaping learning in policy processes are the uncertainty of relevant scientific and technical information, problem definition, and policy alternatives (Maz ur, 1981). Beliefs may also change or be reinforced due to internal factors for example limited cognitive abilities, perceptual filters, or pre existing belief structures which may affect information processing (Simon, 1985). Learning may also be shaped into different forms or products such as belief change and belief reinforcement from interactions and activities between and among allies and individuals from groups that advocate different policy goals ( Sabatier & Jenkins Smith 1993 1999)

PAGE 87

! 73 Specificall y, the extent to which policy actors engage in policy activities such as collaboration, coordination, and consensus building within and between advocacy coalitions may also affect information acquisition and potentially shape learning in the direction of b elief change or reinforcement ( Sabatier & Jenkins Smith 1993 1999 ; Jenkins Smith et al., 2014 ) Better understanding the factors that affect individual policy learning is an important aspect toward gaining a more complete picture of how policy actors learn in the policy process. This chapter uses the Advocacy Coalition Framework (ACF) as a lens to view the processes and products of learning. Factors such as extremeness of beliefs, different policy activities and sources of learning may shape learning in different ways and lead to various forms of learning. Specifically, the role of these variables will be tested to determine effects on individual policy actor policy learning, defined here as including belief change belief reinforcement and combinations of changes and reinforcement This leads to the research question of this chapter: What factors affect policy actor belief change and reinforcement in a local and state level energy and climate policy subsystem? THEORY & HYPOTHESES: POLICY LEARNING AND THE ADVOCACY COALITION FRAMEWORK The Advocacy Coalition Framework (ACF), developed by Sabatier and Jenkins Smith has become a leading framework to examine among other policy processes, individual polic y actor belief and belief change ( Birkland, 2011 ). T he ACF builds from a boundedly rational model of the individual policy actor possessing perceptual filters stemming from belief structures spending time and resources gaining knowledge of

PAGE 88

! 74 specific polic y areas and building political networks in order to affect the development of public policy ( Sabatier, 1988; Jenkins Smith et al., 2014 ). I ndividual policy beliefs 7 defined as a collection of normative and empirical beliefs spanning the substantive policy topic (in this case climate and energy policy) in a geographic area (in this case Colorado), the ACF argues, pull actors with similar beliefs together in to advocacy coalitions ( Sabatier 1988 ; Sabatier & Jenkins Smith, 1993 199 9 ) A dvocacy coaliti ons are group s of different policy actors (e.g. i nterest group representatives, lawmakers, agency officials, or researchers ) who share policy beliefs and, to at least some extent, collaborate in their attempts to influence the policy process ( Jenkins Smith et al., 2014 ). The policy actors as individuals, as opposed to their advocacy coalitions, will be the unit of analysis for this chapter, and broader dissertation study, but comparisons will be made between policy actors with differen t self re ported belief systems. A 2009 comprehensive meta analysis of applications of the ACF found that some studies have shown policy learning and individual belief change within and across coalitions to have potential to drive policy development and change over time (Weible et al ., 2009). But previous ACF applications of learning have been inconsistent in the conceptualization and measurement of this concept and it is considered an aspect of the ACF deserving of greater practical attention and theoretical innovat ion (Jenkins Smith et al., 2014; Weible, 2011). This chapter attempts to follow the advice of Jenkins Smith et al. (2014, p. 205) to improve the ACF and make clear that both belief change and belief reinforcement are conceptually considered policy learning products, delineate and !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! !!!!!!!!!!!!!!! @ As thi s application of the ACF is not examining differences between deep core beliefs, policy core beliefs, and secondary beliefs the phrase "policy beliefs" will be used.

PAGE 89

! 75 measure each clearly, and examine what factors might shape policy learning in either of these directions Theoretical Emphasis The ACF provides a number of causal relationships and hypotheses regarding policy beliefs and learning (for a full list of hypotheses see: Jenkins Smith et al., 2014). There are three major theoretical emphases of the ACF which are typically the focus of applications in the literature: 1) advocacy coalitions (mapping the structure, membership, resources, an d changes over time) ; 2) policy change (examining and explaining the policy developments over time and the role of coalitions therein) ; and 3) the concept of policy learning or policy oriented learning. The third of these theoretical emphases will be the focus of t his chapter and broader dissertation study. To situate this chapter in proper context, a brief discussion of the study of learning in disciplines other than public affairs is necessary. As Muro and Jeffr e y (2008) point out, various and diverse fields such as social psychology and neuroscience have been struggling to understand learning for decades. Despite a great deal of studies it is difficult to make comparisons across much of the work in part because of different assumptions made related t o the nature of learning ( Muro & Jeffr e y 2008) The concept of learning is multidimensional and this can lead to confusion between the concept of learning itself and its potential outcomes for instance (Reed et al., 2010). In other words, is simply acqu iring new knowledge enough to constitute learning, or must a deeper behavioral change result as well to indicate learning has occurred? Studies that assume new knowledge must be incorporated and utilized by individuals to constitute learning are making und erlying assumptions that other

PAGE 90

! 76 researchers do not (Crona & Parker, 2012). For instance, a different approach to the question posed above is to ask if individuals in social situations are capable of learned behavior through conditioning, with no additional information acquisition (Muro & Je ffrey 2008). The learning context and the motivations for learning are also critical factors to consider when compar ing studies. Recent meta studies of learning (see Muro & Je ffrey 2008; Re ed et al., 2010; Crona & Parke r 2012) agree that while no singular definition, conception, or operationalization of learning exists a cross the social sciences learning is often cited as a normative goal and perhaps a prerequisite for adaptive governance. It is therefore important to delineate the different dimensions of learning and the factors that might shape pol icy actor learning. Heclo (1974) is commonly cited in public policy literature as an early and important contributor to the study of learning in policymaking and this work indeed influenced Sabatier's early theorizing of the ACF relating learning to policy change But there is of course diversity in the way learning is conceptualized in the policy process literature. Some make distinctions between cognitive versus behavioral changes (Gerlak & Heikkila, 2011) while others look for organizational changes in collective governance settings (Crona & Parker, 2012). The contextual factors that describe how social environments and networks can cultivate learning are also important t o explore (Reed et al., 2010), as are broader descriptions of the political environment (Gerlak & Heikkila, 2011). Likewise, the role of learning has evolved into a goal, sometimes explicitly in collaborative government literature (Muro & Jeffrey 2008) an d the outcomes of learning may be conceptualized simply as policy change (Busenberg, 2001).

PAGE 91

! 77 The concept of learning is central to many theories, models, and frameworks used in studying public policy and policy processes (Freeman, 2006; Birkland, 2011; Heik kila & Gerlak, 2013). Learning, as a part of the policy process, is particularly relevant within the ACF as evidenced by the use of the word "learning" in the title of the seminal article introducing the ACF (Sabatier, 1988). Within the ACF, learning has been defined as: "relatively enduring alterations of thought or behavioral intentions that result from experiences and/or new information and that are concerned with the attainment or revision of policy objectives" (Sabatier & Jenkins Smith, 1999, p. 123). However, across applications of the ACF, learning has been variously and poorly defined theoretically (Weible et al., 2011). While dozens of applications of ACF have affirmed that opposing advocacy coalitions are formed and maintained based on stable poli cy beliefs among individuals within policy subsystems (Weible et al., 2011; Weible et al., 2009 ; Zafonte & Sabatier, 1998 ), learning has been the least explored of the three major ACF theoretical emphases (Jenkins Smith et al., 2014). One of the reasons fo r this is that there is little agreement on clear conceptualization and operationalization of learning. Thus systematic comparison between different products of learning and the various processes of learning that shape different products has been a difficult and often ignored or avoided endea vor (Heikkila & Gerlak, 2013). In the studies conducted over the previous decades scholars working within the ACF have found policy learning to be more likely in a c ontext where conflict is at interm ediate levels and focuse d on specific policy instruments or variations and when there are forums with established professional norms (e.g. mediated roundtables) for individuals in opposing coalitions to collaborate on policy development ( Lester & Hamilto n, 1987; Sabatier & Brasher, 1993; Eberg 1997; Ellison 1998; Lubell, 2003;

PAGE 92

! 78 Meijerink, 2005; Larsen et al., 2006) While the strength of this past research has been to better understand policy subsystems and collaboratives where policy change may be more likely, a limitation is that policy change was often conflated with the concept of policy learning. Specific models of policy actors learning processes or products have not been tested as often as the larger policy system attributes, such as the level of conflict. One of the central challenges in studying learning, at large and when using the ACF, is a lack of consistency and clarity in how learning is defined not just conceptually, but also operationally. For example, individual beli ef change, coalition change, and policy are variously provided as theoretical indicators of learning in ACF applications (Weible et al., 2009). In the ACF literature, as well as the policy process literature more broadly, one or more of these three indica tors of learning are often provided as evidence of policy learning but rarely is this explicitly linked to a specific conceptualization of policy learning (Heikkila & Gerlak, 2013; Weible et al., 2011). Even fewer attempts have been made to examine polic y learning (even belief change) empirically using survey data; instead, researchers have often relied on assessments of policy change over time from, for example unsystematic content analysis of existing documents or historical analysis (Weible et al., 20 11, 2009). This chapter specifically conceptualizes and defines belief reinforcement in addition to belief change as evidence of learning Reinforcement is another form of cognitive change. Belief reinforcement may take the form for instance, of a more fir mly supported policy belief, more buttressed by new information or experiences. Reported cognitive, as opposed to behavioral, chan ges signify the results (or products) of policy

PAGE 93

! 79 learning for the purposes of this dissertation, as reported in this chapter 8 While both belief change and belief reinforcement may be included in a concept of policy learning in the ACF, this chapter includes the argument that they are different kinds of learning. In other words belief change and belief reinforcement could be co nsidered oppositely charged poles with in the concept of policy learning to use an analogy from Goertz (2006). Policy l earning defined here, includes belief change, belief reinforcement and different combinations of these categories such as policy belief that have been mostly reinforced, with some changes or a balance of change and reinforcement Therefore policy actors reporting neither belief change, nor belief reinforcement are in dicating not learning or non learning in terms of a lack of cognitive change. In other words, no change or reinforcement of beliefs could be evidence of non learning (the absence of cognitive change). Recent meta analysis of learning in public policy have pointed out that the notion of nonlearning is a conceptually difficult area ( Heikkila & Gerlak, 2013; Gerlak & Heikkila 2011 ). In keeping with this study's attempt to be clear on conceptualization and measurement of learning, for the purposes of this chapter, the lack of cognitiv e change or reinf orcement will be considered non learning. Largely untested, according to Jenkins Smith et al. (2014), ACF hypotheses do exist relating policy actors' beliefs to learning. For instance, it is suggested policy actors may display variation in the extremeness o f policy positions they espouse. Further, more extreme (less moderate) views of policy actors regarding specific policy issues may make policy actors less flexible to change in their beliefs related to those issues. The ACF !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! !!!!!!!!!!!!!!! 8 While most of the learning being examined in this dissertation relates to belief change or reinforcement (cognitive changes), individuals were asked about changes in advocacy strategies used to forward policy goals this represents a behavioral change and the investigation into this learning is presented in a later analysis.

PAGE 94

! 80 arguments are based on the largely accepted claim that individual perceptual filters might bias information assimilation in ways that could lead to confirmation bias and belief reinforcement (Lord et al., 1979). These infrequently examined ACF hypotheses h ave some mixed support in the literature (Jenkins Smith et al., 2014) As the objective of this chapter is to further refine ACF hypotheses and models of individual policy learning this chapter will present findings from analyses that test the following h ypothes i s, as adapted fr om Jenkins Smith et al. (2014). Extreme Belief Hypothes i s : H ypothesis 1 : Policy actors with more extreme policy views are more likely to reinforce their beliefs than change their beliefs The ACF argues that policy activities, for example interactions between and among individuals that advocate similar policy goals and/or those that advocate different policy goals may affect policy learning ( Sabatier & Jenkins Smith 1993 1999). For instance the extent to which policy actors engage in policy activities such as seeking advice, collaboration, and consensus building within and between advocacy coalitions may affect information acquisition. Specifically, if policy actors' activities are more ins ular in nature in terms of interacting more exclusively, e.g. seeking advice more frequently from individuals who share their beliefs on climate and energy policy this could potentially sha pe policy learning through mechanisms like confirmation bias (the process of selection of information that confirms existing preconceptions) in the direction of belief reinforcement ( Sabati er & Jenkins Smith, 1993 1999 ). Conversely, policy actors engaging in activities that expos e them to more diverse ideas, for instance those that engage in more frequent collaboration with individuals with

PAGE 95

! 81 dissimilar policy beliefs or objectives, may report policy learning shaped in the direction of belief change. This would be consistent with the finding of Leach et al. (2013), that new information acquired in collaborative processes also has the potential to change policy beliefs. Beyond collaboration with those with dissimilar beliefs the ACF contends that policy actors participat ion in facili tated consensus based process (e.g. mediated roundtables) and constructive cross coalition activities may expose them to diverse ideas on the conceptual proble m and possible policy solutions, and potentially shape learning in the direction of belief chang e. These arguments lead to three policy activity related hypotheses to be tested, also adapted fro m Jenkins Smith et al., (2014). Policy Activities Hypotheses : Hypothesis 2 : Policy actors that seek advice more frequently from those with similar beliefs are more likely to reinforce their beliefs rather than change their beliefs. Hypothesis 3 : Policy actors that collaborat e more frequently with those with dissimilar beliefs are more likely to change their beliefs rather than reinforce their beliefs Hypothesis 4 : Policy actors that have participated in more frequent facilitated consensus based processes are more likely to change their beliefs rather than reinforce their beliefs CLIMATE CHANGE POLICY The Intergovernmental Panel on Climate Change and the broader scientific community unequivocally claim and demonstrate that, without intervention, climate change will have devastating impacts on human communiti es (Giddons, 2011; IPCC, 2013 2014; Stern 2007). Until very recently the federal government in the U.S. has done very little promulgating meaningful climate legislation. This vacuum has provided

PAGE 96

! 82 the space for a surge of policy action at the subnational level, where localities and counties serve as laboratories for the cre ation, implementation, and examination of promising programs and legislation Thus, the study of subnational climate and energy policy is becoming i mportant among political scientists and policy scholars struggling to understand the actors and process es in place in these sub national policy subsystems. Subnational Climate Policy Landscape On February 16 2005 the Kyoto Protocol, the international agreement to address climate disruptio n, went into effect and almost 200 countries have ratified it to date. Despite the fact that the U.S. was a signatory to th e agreement, Congress made it clear it had no intention of ratifying the treaty and thus the U.S would not be participating in the Kyoto Protocol ( Regan, 2015; Layzer 2006). Determined to address th e i ssue, on that same February day the Kyoto Protocol went into effect across the globe, Seattle Mayor Greg Nickels launched the U S Mayors Climate Protection A greement to advance the goals of the Kyoto Protocol through local government leadership and action ( US Conference of Mayors, 2015 ). There are more than 1,000 signatories to the agreement, which is now called the US Conference of Mayo rs Climate Protection A greement (USMCPA) including Denver, Colora do (US Conference of Mayors, 2015 ). Participating cities have agreed to reduce community wide greenhouse gas (GHG) emissions to at least 7 percent below 1990 levels or better by 2012 The signatory cities have commit ted to meet or exceed the Kyoto Protocol GHG emission targets in their own communities by using a litany of climate related polices and plans 9 Climate and energy p olicy at the state level has also !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! !!!!!!!!!!!!!!! 9 For a review of the results of these trends see Krau se, 2011 and Wood et al., 2014.

PAGE 97

! 83 been created in response to the lack of national pol icy. At least 30 U.S. states, including Colorado, and hundreds of U.S cities including Denver, have created a climate action plan (CAP) of some sort ( EPA, 201 5 ). A city or state CAP typically outlines broad policy goals a s well as specific programmatic a nd policy recommendations that will be used as greenhouse gas reduction measures Denver and Colorado Climate Policy Landscape In 2005 then Denver Mayor John Hickenlooper launched Greenprint Denver, a new department in the City and County of Denver created to "advance and further support the integration of environmental impact analysis into the city's programs and policies, alongside economic and social analysis" (City and County of Denver, 2006, p. 2). This new program built on the city 's history of climate policy going back a decade is an early member of the International Council of Local Environmental Initiatives (now know n simply as ICLEI) (Bulkeley & Betsill, 2003). Compared to other medium to large U S cities that are signatories to the USMCPA, Denver is typical in terms of its population density, stated climate policy goals, and current progress in attaining said goals. In November 2007 then Colorado Governor Bill Ritter launched an initiative to address climate change statewide, whi ch resulted in the creation of the Colorado Climate Action Plan. This plan called for a reduction of the state emission of greenhouse gases by 20 percent below 2005 levels by 2020 and a longer term goal of 80 percent below 2005 levels by 2050 (Colorado Cli mate Action Plan: A Strategy to Address Global Warming, 2007). In terms of scale and scope of policies as well as the shorter term emission reduction targets, the Colorado CAP is typical of state CAPs (EPA, 2015; Ramseur, 2007). Given this, Colorado could be thought of as a typical case in terms of case selection (Gerring, 2007).

PAGE 98

! 84 S ince the inception of both the plans Mayor Hickenlooper who spearheaded the Denver level climate action plan has become the Governor of Colorado and is now leading the charge on the state CAP. The policy actors from across Colorado, including those from Denver, will be treated together as subnational actors for the following reasons: Denver is the largest city i n Colorado, the state's capital, and influential in state level policy developments in climate and energy issues, and often, as in this case, state and cit y governments cooperate on energy and climate policy lastly, given the leadership and administrative changes at the local and state level, there is overlap in c ity and state political actors. DATA AND METHODS Data Collection Th is study is an investigation of the climate and energy policy subsystem of Colorado using data collected in 2011 from an original internet based survey The target population was p olicy actors employed by or involved with government agencies, nonprofit organizations, and private companies engaged in climate and energy policy in Colorado This included individuals who attended the roundtable sessions preceding the creation of Greenprint Denver and the Col o r ado CAP, and the advisory council members of both plans A snowball sample was generated from this set of individuals. The survey was sent to 793 poli cy actors involved in climate and energy issues in Colorado and 260 individuals returned fully completed surveys (response rate of 33 percent ) The sampling technique employed for the survey is explicitly purposive and non probability based because the vie ws and opinion of active policy actors is ex plicitly the goal of this study

PAGE 99

! 85 Data Analysis The quantitative data collected with the surveys were analyzed using STATA 13 software. Descriptive statistics and crosstabs were calculated. Given the nature of the dependent variable as non ordinal scale of policy learning, multinomial logistic regressions w ere used to test the effect of extreme beliefs and policy activities on belief change or reinforcement. The normality, skewedness, and relevant assumptions regarding the tests were examined, as well as the potential for any interactions between in dependent variables. Opera tional Measures Table 3.1 below presents the operational measures used for each of the independent variables and the dependent variable. In other words, the exact survey questions used to measure each of the variables in this chapter, as well as the format of all possible responses for each question, are presented below. For responses that were coded for statistical purposes, that information is also provided for each question. For sake of brevity, this is not a comprehensive li st of all the survey question s. Each variable concept is listed in bold, and below each concept is the operational definition and survey question, used to measure each variable. T hree different policy activities measured the frequency with which the policy actors engaged in each activ ity on a 5 point frequency scale. Extremeness of policy beliefs (independent variable for hypothesis 2) w as measured using a 5 point Likert scale of agree/disagree for six policy belief statements. T he dependent variable individual policy learning was measured for each of the six policy belief questions (in other words, policy learning was measured six

PAGE 100

! 86 times, once for each of the six policy beliefs), and six categories of responses were offered regarding the form of policy learning. Table 3.1 Operation al Measures of Attributes Policy Activities Respondents were asked to report how often they participated in the following activities as part of their work in the past year (on a scale of: d aily, w eekly, m onthly, y early, and n ever) : Collaboration with those who do not share their views on energy & climate goals Seek a dvice from individuals the y agree with Facilitated consensus building processes (e.g. focus groups, roundtables) These were coded in the following way: Never = 0 Year ly = 1 Monthly = 2 Weekly = 3 Daily = 4 Policy Beliefs & Extremeness of Belief Respondents were asked to indicate their rel ative agreement (on a scale of s trongly agree, s omewhat agree, n either agree nor disagree, s omewhat disagree, and s trongly disagree) with the following stat ements: The severity of predi cted impacts on society from climate change are vastly overstated Human behavior is the principal cause of climate change Decisions about energy and its effect on climate are best left to the e conomic market, and not to the government An energy and/or carbon tax is required to combat climate change A cap and trade system of permits for the emission of greenhouse gasses is required to combat climate change Government policies to promote renewab le energy generation are required to combat climate change Extremeness of Belief was created by coding answers for each policy belief question in the following way: S trongly agree and s trongly disagree = 2 Somewhat agree and s omewhat disagree = 1 Neither agree nor disagree = 0 Operational Measure s / Survey Question s: Learning For each of the six policy belief questions above, respondents were asked to what extent their views had changed or been reinforced (no time frame was associated with these questions): Only been reinforced Mostly reinforced with little changes Balance of reinforcement and changes Mostly changed with little reinforcement Only been changed Neither changed nor been reinforced

PAGE 101

! 87 RESULTS AND ANALYSIS Descriptive Statistics Descriptive statistics for eac h variable are presented below. Extreme Beliefs Table 3.2 below illustrates that relatively extreme beliefs were the norm in four of the six policy beliefs : the mean extreme belief score ranged between 1.45 and 1.48, where 2 would indicate every individual reported an extreme score. The policy beliefs regarding t he need for an energy and/or carbon tax and the need for a cap and trade system had lower mean ext reme belief scores than the trend seen in the other four policy beliefs Table 3.2 Extrem eness of Policy Beliefs Policy Belief Mean Extreme Belief Score (Range 0 2) Strongly Agree & Strongly Disagree (Coded as 2) Somewhat Agree & Somewhat Disagree (Coded as 1) Neither Agree nor Disagree (Coded as 0) The severity of predicted impacts on society from clim ate change are vastly overstated 1.48 152 (58%) 1 82 (32%) 1 26 (10%) Human behavior is the principal cause of climate change 1.47 141 (54%) 1 99 (38%) 20 (8%) Decisions about energy and its effect on climate are best left to the economic market, and not to the government 1.45 135 (52%) 106 (41%) 19 (7%) An energy and/or carbon tax is required to combat climate change 1.23 101 (39%) 119 (46%) 1 40 (15%) A cap and trade system of permits for the emission of greenhouse gasses is required to combat climate change 1.02 1 71 (27%) 124 (48%) 1 65 (25%) Government policies to promote renewable energy generation are required to combat climate change 1.45 165 (63%) 1 87 (33%) 1 8 (3%)

PAGE 102

! 88 Policy Activities As can be seen in Figure 3.1 below, r espondents reported they sought advice from individuals they agreed with more frequently than they collaborated with individuals with di ssimilar views on climate and energy goals. Participation in facilitation/consensus building was overall less common. In fact 159 of 260 respondents (61 percent) reported they engaged in facilitation/consensus building (such as focus groups and roundtables) only monthly or yearly and 49 respondents (19 percent) reported they never engage in this policy activity. Figure 3.1 Frequency of Policy Activities "! '"! #"! ="! $"! >"! %"! @"! &"! A"! '""! O/-0+!! Y))Z0+! 1264C0+! W)/*0+!! H)8)*! !"#$%&'()'*+,--.-,"/01' B&%F"%+83' B&%F"%+83'()'C+/:%#%+6'-+'7(0-83' <86-.-6-%1' <))Z!E38-:)!S*2F!a63-8-3M/07!a!ET*))!Y-4C! 5200/.2*/4)![-4C!\C27)!YC2!O2!H24!
PAGE 103

! 89 Policy Learning The dependent variable in this chapter was policy learning, defined as a scale between reinforcement of beliefs and change in bel iefs. Table 3. 3 below shows in terms of forms policy learning, belief reinforcement, as opposed to belief change, was reported with much higher frequency across all six major policy core questions. Reports of neither change nor reinforcement of beliefs (ranging from 6.5 27.2 percent of responses) were also reported with higher frequency than belief change. Table 3. 3 Policy Learning Across the S ix Policy Belief Questions Policy Belief a nd Learning My Views Have Only Been Reinforced My Views Have Been Mostly Reinforced With Little Changes My Views Have Had a Balance of Reinforcement & Changes My Views Have Mostly Changed With Little Reinforcement My Views Have Changed Only My Views Have Neither Changed Nor Been Reinforced The severity of predicted impacts on society from climate change are vastly overstated 92 (35.4%) 87 (33.5%) 62 (23.8%) 2 (<0.1%) 0 (0.0%) 17 (6.5%) Human behavior is the principal cause of climate change 104 (40.3%) 76 (29.5%) 51 (19.8%) 1 (<0.1%) 1 (<0.1%) 25 (10%) Decisions about energy and its effect on climate are best left to the economic market, and not to the government 66 (26.0%) 66 (26.0%) 70 (27.1%) 4 (1.6%) 4 (1.6%) 48 (18.6%) An energy and/or carbon tax is required to combat climate change 44 (17.2%) 83 (32.4%) 67 (26.2%) 7 (2.7%) 3 (1.2%) 52 (20.3%) A cap and trade system of permits for the emission of greenhouse gasses is required to combat climate change 37 (14.6%) 58 (22.8%) 72 (28.3%) 12 (4.7%) 6 (2.4%) 69 (27.2%) Government policies to promote renewable energy generation are required to combat climate change 111 (43.5%) 56 (22.0%) 34 (13.3%) 3 (1.1%) 3 (1.1%) 48 (18.8%)

PAGE 104

! 90 Testing Hypotheses 1 4 : Extreme Beliefs and Policy Activities Effect on Learning : Multinomial Logistic Regressions The dependent variable for all hypotheses in this chapter is policy learning, conceptualized as belief change and belief reinforcement and variations of these categorical forms of learning. In terms of how the learning concept was operationalized, this was presented as a 5 point scale including belief change, belief reinforcement, and combinations of these two as well as a sixth possible response of my views have neither changed nor been reinforced For this reason, the dependent variable cannot be treated as ordinal 10 In these cases the multinomial logit model (MNLM) also known as multinomial logistic regression (MNLR) is the most frequently used model (Long & Freese 2014). Thus multinomial logistic regression analysis was used to test how the various in dependent variables were associated with the probability of different categories of learning ; i n other words, testing how the independent variables affected the likelihood of observing different products of learning across the continuum of belief change an d belief reinforcement for each of the policy beliefs. The MNLM can be thought of as a set of binary logits among all pair comparisons of alternative levels of the dependent variable, in this case the six levels of learning (Long & Freese, 2014; Long, 1997 ). Six MNLRs, one for each policy beliefs learning question, were run using the same model and the base outcome was set to "belief reinforcement only" in all analyses. The most effective methods of interpretation of the MNLM are those that help understand predicted probabilities, and specifically changes to those probabilities in the level of the dependent !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! !!!!!!!!!!!!!!! 10 This is similar to how offering a "do not know" option as a sixth option to a 5 point Likert agree/disagree question would invalidate the ordinal outcome models.

PAGE 105

! 91 variable based on marginal changes in the independent variables (Long & Freese, 2014). The marginal change is essentially the average percent change acro ss the dependent variable (the probability of change) associated with changes to each independent variable, holding all other independent variables constant. After each of the six MNLR is run a marginal change (mchange) plot is calculated which tests the likelihood of effects in the dependent variable caused by one increase in standard deviation of each independent variable. The raw outputs of each of these marginal change analyses are dense and difficult to interpret intuitively and thus are p resented in Appendix A of this dissertation. Marginal change plots presented below are much easier to interpret. In marginal change plots the average discrete change in the dependent variable based on one standard deviation increase in each independent var iables (holding all others constant) are plotted together. Each possible outcome or level of learning is represented by a single letter: R = Reinforcement Only, r = Mostly reinforcement, B = Balance, c = Mostly Changed, C = Changed only, and N = No Change or Reinforcement. The resulting changes in the dependent variable are represented by shifts, either more likely or less likely (to the right or to the left respectively from the center line) based on the effects of the independent variables. These plots t herefore represent changes in pr edicted probability. Asterisks (*) indicate a significant relationship ( p < 0.05) between a change in the range of an independent variable and a change in the predicted probability between two categories of the dependent variable. M arginal change plots for each of the six policy belief questions are presented below. M easure of fi t for the learning model in each of the plots are pr ovided with the Pseudo R 2 specifically McFadden's R 2

PAGE 106

! 92 also known as the LR index the most common measure of fit presented in a n MNLR (Long & Freese 2014) 11 The complete multinomial logit raw output with belief reinforcement only (R) as the baseline can be found in Appendix A. Collinearity diagnostics were performed for each of the six MNLR and the tolerance and variance inflation factors (VIFs) indicated no multicolinearity issues in any of the MNLRs. Tab le 3.5 below shows the independent variable labels that were used for the sake of parsimony in the following MNLR marginal change plots. To facilitate interpretation of the marginal change plots, the first of six (Figure 3.2) will be explicated. Each of th e independent variables appear as rows in the marginal change plots, listed on the left hand side. The six possible outcomes of learning are indicated with one letter each and labeled in the charts. If in the statistical analysis, the probability of one outcome of learning (the dependent variable) is affected by one standard deviation increase of any independent variable, then the letters representing that outcome of learning will shift either to the left of the center line (indicat ing it is less likely) or to the right of the center line (indicating it is more likely). If that change in probability of outcome is significantly affected ( p < 0.05), either more likely or less likely, it will appear with an (*) indicating that change in probability is significantly outside the predicted range. If there is no (*) associated with the letter indicating level of the dependent variable (outcome of learning) than the chang e was within predicted ranges. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! !!!!!!!!!!!!!!! 11 Because this statisti c does not mean exactly what R 2 means in OLS regression (the proportion of variance in the dependent variable predicted by the independent variables), it should be interpreted with some caution.

PAGE 107

! 93 Table 3. 4 Independent Variable Labels in Marginal Change Plots Independent Variable MNLR Label Respondents were asked to indicate their relative agreement (on a Likert scale of s trongly agree, s omewhat agree, n either agree nor disagree, s omewhat disagree, and s trongly disagree) wit h six policy belief statements. Extremeness of Belief was created by coding answers in the following way: Strongly agree and s trongly disagree = 2 Somewhat agree and s omewhat disagree = 1 Neither agree nor disagree = 0 Extreme ness of Belief * There were six different policy belief questions, providing the opportunity to test the model across six different dependent variable measures of policy learning. The extremeness of b elief score was used as an independent variable with each of the corresponding policy learning question s, the dependent variable. Respondents were asked to report how often they participated in the following activities as part of their work in the past year (on a scale of: d aily, w eekly, m onthly, y early, and n ever): Se ek advice from individuals they agree with Frequency of See king Advice from Individuals I A gree with Collaboration with those who do not share their views on energy & climate goals Frequency of Collaborating w ith Individuals I Disagree with Facilitated consensus building processes (e.g. focus groups, roundtables) Frequency of Participation in Facilitated Consensus B uilding Processes

PAGE 108

! 94 Figure 3.2 Changes in Predicted Probabilities in Learning Regarding the Severity of Predicted Impacts of Climate Change R = Reinforcement Only, r = Mostly reinforcement, B = Balance, c = Mostly Changed, C = Changed only, N = No Change or Reinforcement Pseudo R 2 = 0. 17 and the number of observations = 251 (*) indicates a significant relationship (p < 0.05) btw a cha nge in the range of an IV and change in predicted probability btw two categories of the DV R* r* B* c N* R* r* B c N R r B c N R r B c N Extremeness Frequency of Seeking Advice from Individuals I Agree with Frequency of Collaborating with Individuals I Disagree with Frequency of Participation in Facilitated Consensus-Building Processes of Belief .15 -.25 -.2 -.15 -.1 -.05 .05 .1 .2 .25 Marginal Effect on Outcome Probability

PAGE 109

! 95 Figure 3.3 Changes in Predicted Probabilities in Learning Regarding Humans as the Cause of Climate Change R = Reinforcement Only, r = Mostly reinforcement, B = Balance, c = Mostly Changed, C = Changed only, N = No Change or Reinforcement Pseudo R 2 = 0. 16 and the number of observations = 249 (*) indicates a significant relationship (p < 0.05) btw a change in the range of an IV and change in predicted probability btw two categories of the DV R* r B* c N R R r B c N R R r B c N R R r B c N R Extremeness of Belief Frequency of Seeking Advice from Individuals I Agree with Frequency of Collaborating with Individuals I Disagree with Frequency of Participation in Facilitated Consensus-Building Processes -.2 -.15 -.1 -.05 .05 .1 .15 .2 -.25 .25 Marginal Effect on Outcome Probability

PAGE 110

! 96 Figure 3.4 Changes in Predicted Probabilities in Learning Regarding the Need for Government to Address Climate Change R = Reinforcement Only, r = Mostly reinforcement, B = Balance, c = Mostly Changed, C = Changed only, N = No Change or Reinforcement Pseudo R 2 = 0. 0 8 and the number of observations = 249 (*) indicates a significant relationship (p < 0.05) btw a change in the range of an IV and change in predicted probability btw two categories of the DV R* r B* c C N* R r B c C N R r B c C N R r B c C N Extremeness of Belief Frequency of Seeking Advice from Individuals I Agree with Frequency of Collaborating with Individuals I Disagree with Frequency of Participation in Facilitated Consensus-Building Processes .05 -.25 -.2 -.15 -.1 -.05 .1 .15 .2 .25 Marginal Effect on Outcome Probability

PAGE 111

! 97 Figure 3.5 Changes in Predicted Probabilities in Learning Regarding the Need for a Carbon Tax to Address Climate Change R = Reinforcement Only, r = Mostly reinforcement, B = Balance, c = Mostly Changed, C = Changed only, N = No Change or Reinforcement Pseudo R 2 = 0. 06 and the number of observations = 249 (*) indicates a significant relationship (p < 0.05) btw a change in the range of an IV and change in predicted probability btw two categories of the DV R r* B c C N* R r B c C N R r* B c C N* R r B c C N Extremeness of Belief Frequency of Seeking Advice from Individuals I Agree with Frequency of Collaborating with Individuals I Disagree with Facilitated Consensus-Building Processes Frequency of Participation in -.15 -.1 -.05 .05 .1 .15 Marginal Effect on Outcome Probability

PAGE 112

! 98 Figure 3.6 Changes in Predicted Probabilities in Learning Regarding the Need for a Cap & Trade System R = Reinforcement Only, r = Mostly reinforcement, B = Balance, c = Mostly Changed, C = Changed only, N = No Change or Reinforcement Pseudo R 2 = 0. 08 and the number of observations = 246 (*) indicates a significant relationship (p < 0.05) btw a change in the range of an IV and change in predicted probability btw two categories of the DV R* r* B* c C N* R r B c C N R r B c C N R r B c C N Extremeness of Belief Frequency of Seeking Advice from Individuals I Agree with Frequency of Collaborating with Individuals I Disagree with Frequency of Participation in Facilitated Consensus-Building Processes -.1 -.15 -.05 .05 .1 .15 Marginal Effect on Outcome Probability

PAGE 113

! 99 Figure 3.7 Changes in Predicted Probabilities in Learning Regarding the Need for Policies Promoting Renewables R = Reinforcement Only, r = Mostly reinforcement, B = Balance, c = Mostly Changed, C = Changed only, N = No Change or Reinforcement Pseudo R 2 = 0. 16 and the number of observations = 247 (*) indicates a significant relationship (p < 0.05) btw a change in the range of an IV and change in predicted probability btw two categories of the DV R* r* B* c C N* R r B c C N R r* B c C N R r B c C N Extremeness of Belief Frequency of Seeking Advice from Individuals I Agree with Frequency of Collaborating with Individuals I Disagree with Frequency of Participation in Facilitated Consensus-Building Processes -.1 .25 .2 .15 .1 .05 -.05 -.15 Marginal Effect on Outcome Probability

PAGE 114

! 100 Extreme Beliefs 12 The extremeness of policy actors' beliefs had a powerful effect on their policy learning across all six policy beliefs. One standard deviation increase in this independent variable led to three different statistically significant and relatively consistent trends in the changes to the dependent variable across all six policy learning q uestions. First, one standard deviation increase in extreme views (more extreme views) led to policy actors being less likely to report a balance of belief change and reinforcement in five of the six policy beliefs: 12 percent less likely regarding the pro jected severity of climate change impacts 13 percent less likely regarding humans as the primary cause of climate change, 10 percent regarding the need for government intervention to address climate change, 8 percent less likely regarding the need for a c ap and trade system, and 5 percent less likely regarding the need for policies to promote renewable energy. These changes in predicted probabilities were all statistically significant ( p < 0.05). A second set of significant and corresponding changes in pol icy learning resulted from one standard deviation increase in the extreme views of policy actors and their policy learning. More extreme views led policy actors to be much more likely to report !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! !!!!!!!!!!!!!!! 12 Note: To test model sensitivity, the government/non government organization affiliation variable, as well as the full 5 option organization affiliation were included in earlier versions of the MNLR model in neither case did organization affiliation signifi cantly affect learning or did it effect the results presented in the model used. Based on existing ACF hypotheses essentially relating organizational affiliation and extreme views, a Kruskal Wallis test was used to test whether policy actors from governmen t agencies reported more moderate (less extreme) beliefs than policy actors from ideological/purposive groups or material groups the results do not support this hypothesis. Policy actors from government agencies reported no significant difference in extrem eness of beliefs compared to non government policy actors. A Pearson's Chi square test was also conducted to see if there was any significant difference in belief extremeness across any of the five organizational affiliations, and the results were the same there was no significant relationship between organizational affiliation and belief extremeness.

PAGE 115

! 101 belief reinforcement only in five of the six policy beliefs: 2 5 percent more likely regarding the projected severity of climate change impacts 22 percent regarding humans as the primary cause of climate change, 20 percent more likely regarding the need for government intervention to address climate change, 13 percent regarding the need for a cap and trade system to address climate change, and 25 percent more likely regarding the need for policies to promote renewable energy. Similarly, one standard deviation increase in the extreme views of policy actors lead to policy actors reporting mostly reinforced beliefs in two policy beliefs: 12 percent more likely regarding the need for a carbon tax to address climate change and 8 percent more likely regarding the need for a cap and trade system. This trend of associat ion between extreme views and belief reinforcement was further supported by two additional related significant findings; one standard deviation increase in extreme beliefs led to increased reports of mostly belief reinforcement regarding two policy beliefs : 12 percent regarding the need for a carbon tax, and 7 percent regarding the need for a cap and trade system. These changes in predicted probabilities were also all statistically significant ( p < 0.05) based on the MNLRs. The third trend observed in the e ffects of extreme beliefs on policy learning is that one standard deviation increase in this variable was associated with policy actors being 5 10 percent less likely to report neither change nor reinforce ment of beliefs in five of the six policy belief questions, these were all statistically significant ( p < 0.05) changes in predicted probabilities. The only policy belief this trend was not observed in was regarding humans as the primary cause of climate change. One standard deviation increase in the ext reme views of policy actors also led to two statistically significant ( p < 0.05) changes in predicted probabilities in levels of

PAGE 116

! 102 policy learning that are at odds with the trends above. More extreme views led to 7 percent less likely reports of mostly belie f reinforcement regarding the predicted severity of climate change impacts, and 7 percent less likely reports of mostly belief reinforcement regarding the need for policies to promote renewable energy. It is worth noting these two contradictory results hav e smaller effect size s inconsistent across the six beliefs, and represent just two points in the overwhelming three trends presented above regarding the relationship between policy actors' extrem e views and learning. Frequency of Seeking Advice from Indiv iduals with Similar Views The frequency with which policy actors seek advice from individuals with similar views had some statistically significant ( p < 0.05) changes in predicted probabilities in levels of policy learning in only one of the six policy learning questions, and the results were contradictory. One standard deviation increase in this variable made reports of mostly reinforced beliefs 10 percent less likely but reports of belief reinforcement only 10 percent more likely regarding the severity of predicted effects of climate change. No other significant changes in predicted probabilities results from changes in this independent variable across any of the other policy belief learnin g questions. Frequency of Collaborating with Individual s with Dissimilar Views For the third independent variable in the model, the frequency with which policy actors reporting more collaboration with individuals they dis agree with on climate change and energy policy issues, there were some significant trends in the effects on policy learning. First, one standard deviation increase in this variable led to statistically significant decreased levels of belief reinforcement regarding two of the six policy beliefs. Specifically, policy actors were less likely to re port mostly reinforced views : by 6 percent

PAGE 117

! 103 regarding the need for a carbon tax to address climate change, and 6 percent regarding the need for government policies to promote renewable energy. Secondly, one standard deviation increase in the frequency of co llaboration with individuals with dissimilar views also led to statistically significant increases in the likelihood policy actors would report neither change nor reinforce ment of beliefs: 7 percent regarding the need for a carbon tax to address climate ch ange. These changes in predicted probabilities were all statistically significant ( p < 0.05) based on the MNLRs but it is worth mentioning that the effect size (the average percent changes in predicted probabilities) was much smaller than the effects of extreme beliefs. Frequency of Participation in Facilit ated Consensus Based Processes The frequency of policy actor participation in facilitated consensus based processes (e.g. focus groups, roundtables) had no effect on policy learning. No statistically s ignificant changes in predicted probabilities in learning occurred with any change s to this independent variable. Summary of Results Support for Hypotheses The results support H ypothes i s 1 ; more extreme views were associated policy actors being more likely to report policy learning (less likely to report non learning). That learning associated with more extreme views was more likely in the form of belief reinforcement and less likely in the form of a balance of belief ch ange and belief reinforcement. The results indicate little and mixed support for three policy activities hypotheses. Hypothes i s 2 regarding frequency with which policy actors seek advice from individua ls with similar beliefs (as opposed to individuals across belief systems) associated with

PAGE 118

! 104 learning as belief reinforcement was not supported by the results there were contradictory significant effects on policy learning as belief reinforcement or mostly r einforcement with little change in only one of six policy belief questions. Hypothes i s 3 postulated that i ndividuals that report collaboration with those with dissimilar beliefs (rather than those with similar beliefs within belief systems) would more frequently report policy learning as belief change. Support for this hypothesis was mixed. More frequent collaboration with those with dissimilar beliefs was associated with policy actors being less likely to report mostly belief reinforcement in two of six policy belief questions, but also was associated with policy actors being more likely to report no policy learning. Hypoth esis 4 was not support ed by the results policy actor participation in facilitated consensus based processes had no effect on policy learning. DISCUSSION The purpose of this chapter was to better understand the factors that shape individual policy learning in the policy process. Different factors were hypothesized to shape policy actor learning differently, in other words, lead t o different forms of learning. The results indicate there are different processes and products of learning 13 There is strong support for H ypothesis 1 The extremeness of beliefs was positively associated with belief reinforcement, as opposed to belief change. This !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! !!!!!!!!!!!!!!! 13 Note: This chapter informally tested the hypothesis that policy actors from government agencies will report more moderate (less extreme) beliefs than policy actors from ideological/purposive groups or material groups, which is essentially embedded in established but largely untested, hypotheses of the ACF ( Jenkins Smith et al., 2014) Multiple tests in this analysis show organizational affiliation does not a ffect the extremeness of policy actors' beliefs Specifically, these findings do not support existing ACF coalition hypotheses related to administrative agencies (government employees) advocating more moderate positions than interest group allies and purpo sive groups being more constrained in the expression of their beliefs compared to material groups respectively (Sabatier & Jenkins Smith 1993 1999; Jenkins Smith et al., 2014).

PAGE 119

! 105 corroborates previous ACF findings and the general ACF claim that more extreme views may lead to less belief flexibility, and thus allow for more belief change as opposed to belief reinforcement (Lubell 2003; Jenkins Smith et al., 2014). Given the novel way policy learning was operationalized in this study there are additional insights regarding the products of learning (as shaped by extreme beliefs). Namely, that extremeness of beliefs were also negatively associated with reports of a balance of belief change and reinforcement and also negatively associated with reports of neither belief change nor reinforcement In other words, having extre me beliefs made policy learning more likely in policy actors (because reporting no change and no reinforcement of beliefs was conceptualized and operationalized as not learning) and that policy learning was much more likely to be mostly, or only, belief re inforcement and much less likely to be a balance with belief changes. Conversely, policy actors with more moderate (less extreme) policy belief would seem to be less likely to report belief reinforcement than policy actors with more extreme views. This doe s not mean however belief change is more likely, just that reinforcement of beliefs is less likely to be the outcome of policy learning for tho se with more moderate beliefs. There was no support for H ypothes i s 2 that policy actors that more frequently seek advice from individuals with similar beliefs were more likely to report policy learning as belief reinforcement, as opposed to belief change. But, the results provide some mixed support for H ypothes i s 3 the greater frequency with which policy actors r eported they collaborated with those with dissimilar beliefs reported more belief change which again supports established argument and previous findings in the ACF

PAGE 120

! 106 literature ( Sabatier 1988; Sabatier & Jenkins Smith 1993 ; Leach et al., 2013 ) More collab oration with individuals with dissimilar views also led to more reports of neither belief change nor reinforcement but the effect size was small and it was only observed in two policy learning questions. Perhaps the results indicate that some policy actor s may not learn at all from cross coalition collaboration, but some will report policy learning as less belief reinforcement. These results of the advice seeking and collaboration with individuals with dissimilar variable effects collectively show mixed support for ACF hypotheses regarding the possibility of increased debate across coalitions, or in this case individuals who disagree on energy and climate policy issues, leading to more flexibility, or in this case belief ch ange as opposed to reinforcement (Jenkins Smith et al., 2014). Specifically, results here show while increased cross coalition collaboration increased the likelihood of belief change or flexibility, increased cross coalition advice seeking did not show the same effect. These mixed results regarding the effect of policy activities on policy learning were not bolstered when policy actors were directly asked about their participation in facilitated consensus based processes. N o support was found for H ypothes i s 4 participation in facilitated consensus based process was not found to be significant in shaping policy learning. This would seem to contradict the established arguments and hypothes i s in the ACF literature that learning is more likely when a professiona l forum dominated by norms exists ( Sabatier & Jenkins Smith 1993 ; Jenkins Smith et al., 2014) It is important to acknowledge data limitations when drawing conclusions from this study. First, g iven this is a case study of the climate policy subsystem in one state

PAGE 121

! 107 there are some threats to the ability to generalize potential findings to all subnational entities. That said, the explicit goal of th is study was to examine some factors that shape belief change and reinforcement. Therefore it is designed to e xamine specific relationships between variables as opposed to broad comparative observations. A higher degree of internal validity in the relationships between variables in this case is the trade off ( Gerring 2007 ; George & Bennett 2005) Potential case selection effect is mitigated by the fact that Colorado represents a typical case in that Denver (the most populous city and capital of Colorado) is a member of the US Conference of Mayors Climate Agreement. Also both Denver and the State of Colorado h ave state d typical climate goals and made typical progress ( EPA, 2015 ) Second, there may also be diversity in policy actors that was not captured in this chapter or broader dissertation F or instance this analysis considered all government levels ( e.g. state and city) and agencies together. Relatedly, given that Denver is the capital city (higher population and greater connectivity to state level government issues) the total number of local government actors is skewed toward the City and County of Denv er employees, and less so on small communities and other cities in Colorado. Further research is required to unpack the difference between different sub populations of government employees, and a similar argument could be made for the business sector, acad emic/researchers, and non profit employees. Moreover like all self reported opinion based research, the results represent reports potentially limited or biased by sampling design or the answers provided.

PAGE 122

! 108 CONCLUSION The goal of this chapter was to refine models and hypotheses regarding individual policy learning in the policy process. To that end, this is one of the first ACF applications to specifically ask policy actors to report policy learning as belief change and reinforcement and ask about the sourc e of that learning. Given the way learning was conceptualized and operationalized in this chapter, the results tell an especially compelling story regarding one of the four independent variables tested for effects on policy learning (five variables including organizational affiliation which was examined, but not included in the final model) P olicy actors who reported more extreme views were significantly more likely to report belief reinforcement only or mostly belief reinforcement across policy b eliefs, and less likely to report a balance of belief reinforcement and belief change across and neither belief change or reinforcement. Policy l earning includes belief reinforcement, so having extreme views does not impede individual policy learning in fa ct, by asking about the relative incidence of no belief change or belief reinforcement (not lea r ning) this chapter shows extreme beliefs promote policy learning, and specifically policy learning in the form of belief reinforcement. This is perhaps the maj or take away from this study. In terms of products of policy learning, extreme beliefs are associated with belief reinforcement. T he results of this chapter confirm some classic findings; existing beliefs can shape future learning (Lord et al., 1979). Thes e results also potentially help further refine theories of policy actor learning by supporting the notion that, depending on specific factors involved in the learning process, belief reinforcement is a potential product of policy learning in the policy pro cess (Heikkila & Gerlak 2013; Crossan et al., 1999).

PAGE 123

! 109 This chapter presented findings from analyses aimed to address gaps in the ACF literature in at least two ways. First by examining largely untested ACF hypotheses regarding extreme views shaping policy learning (Jenkins Smith et al., 2014, p. 198) and s econd, by using the ACF to explicitly examine learning as multidimensional, that is to include belief change and belief reinforcement as products of policy learning, measure both, and measu re the incidence of neither belief change or belief reinforcement. In doing so, hopefully th e results can inform improvements upon the ACF, and learning related hypotheses specifically due to the fact that it attempted to emphasize a clear conceptualizati on and operationalization of different products of policy learning. Specifically, while previous recent ACF applications have examined the role of learning in coalition behavior and broader policy change (Nohrstedt 2008; Larson et al. 2006; Nohrstedt 20 05) this study hopes to improve ACF theories regarding factors shaping policy actor learning with its focus on the policy actor attributes and stimuli that encourage and shape individual learning Building on actor based theories of learning in the policy process (Heclo, 1974) and the model of individuals as boundedly r ational (Simon, 1957) scholars have developed and refined models of individual learning in policymaking (Heikkila & Gerlak, 2013). F indings here support some, but do not support other, estab lished ACF hypotheses and so improving these specific models of individual learning is desirable in an attempt to better understand the way policy actors learn. Beyond the ACF, scholars using other policy process theories and frameworks may find lessons h ere. For instance, t he role of extreme beliefs as promoting learning and shaping learning toward belief reinforcement (as opposed to belief change), may be instructional to scholars using the Diffusion of

PAGE 124

! 110 Innovation Framework to examine the role of govern ment officials learning from each other (Berry & Berry 2014; Shipan & Volden 2006 2008). Given that learning is often defined in the policy process literature as a prerequisite for collaborative and adaptive governance and effective policymaking (Henry, 2009; Freeman, 2006), though not the only avenue for policy change, further analysis is merited. One lesson learned in this study is that by conceptualizing learning as change and/or reinforcement, and lack of change and/or reinforcement as non learning, the nuance of extreme belief s being associated with less non learning, in other words extreme beliefs promoting learning (as belief reinforcement) was teased o ut. A n alternative approach would be to directly ask survey respondents to provide the direction of their learning if in fact their beliefs were changed. In other words, if policy actors reported belief change, it would be beneficial to understand if that change took the form of moderation, or of increased intensity. In this study, given the way the learning questions were operationalized, this specificity could not be measured. It can be inferred from the survey question provided that reinforcement is different from change, but is it not change ? Given the way the question was provided to survey respondents, the answer is yes, but that may not be the only way to consider reinforcement of beliefs One way to conceptualize belief reinforcement is simply a belief that does not change in valence (tow ard moderation or toward more intensity) but is simply more a firmly held belief that is now more supported by new information or experience In this sense, information and experience builds up like sediment to further anchor an unchanged belief, and this conceptualization is essentially what was used in this study Another conceptualization of belief reinforcement might be movement further toward the extreme

PAGE 125

! 111 end of a belief over time. This would require a pre and post experience question, trying to deter mine a before and after belief, which might result in belief movement toward reinforcement regarding intensity. This is an issue that can be addressed with future studies attempting to further unpack belief reinforcement as a product of learnin g. Future st udies could also attempt to tease out potential difference between different sub populations of policy actors ( e.g. state versus local government employees ) A lso future research c ould attempt to examine how factors promoting and shaping individual policy actor learning found here might affect collective learning processes. A nother lesson this study offers is that the source of policy learning was not directly asked of the survey respondents in a useful way. Explori ng the role of neutral parties, compared to policy actors with similar beliefs or individu als across belief systems as the source of policy learning might lend insights useful to the ACF regarding policy brokers as factors important in policy learning ( Ingold & Varone 2012; Sabatier & Jenkins Smith, 1993; Mintrom & Vergari, 1996). Similarly this exploration could inform policy entrepreneur theories of public policy ( Ingold & Varone 2012; Mintrom & Vergari, 1996; Schneider & Teske 1993; Teske 1992) as well as scholars employing the Multiple Streams to explain the role of entrepreneurial actor centered theories of policy process (Kingdon, 2003; Zahariadis, 2014; Mintrom & Norman 2009 ). This chapter focused on policy learning as cognitive change or reinforcements (or lack there of) and future studies could also examine the role of factors shaping behavioral change. The results presented in this chapter highlight factors that affect individual policy learning. The primary finding is that extreme be liefs are associated with policy learning, and specifically associated with belief reinforcement as the outcome of learning. In other

PAGE 126

! 112 words, policy actors with extreme beliefs are thus more likely to have their policy views reinforced, that is better suppo rted, as opposed to changed. Given the way learning was conceptualized and measured i n this study, that does not mean having extreme beliefs means policy actors' belief will only get more extreme it means having extreme beliefs, as compared to moderate bel iefs, means policy actors will be more likely to learn (less likely to not learn). Perhaps more interestingly, is that the learning outcome is also more likely to be belief reinforcement This suggest s extreme beliefs have an anchoring effect on policy act ors views, and thus these policy actors' beliefs are then only more likely to be further buttressed by experience, as opposed to changed.

PAGE 127

! 113 CHAPTER IV FACTORS SHAPING POLITICAL LEARNING : A STUDY OF POLICY ACTORS IN SUBNATIONAL CLIMATE AND ENERGY ISSUES CHAPTER SUMMARY Policy actors' beliefs and behaviors may be altered by acquired information and experiences in the policy process. This has been defined as learning and is a n important component of many policy process theories and frameworks. One form of this learning is political learning, such as the change or reinforcement of individual policy actors' advocacy strategies. This chapter attempts to contribute to the conceptual understanding and methodology of studying individual policy actor learning in t he policy process by explicitly examining change and reinforcement of advocacy strategies as behavioral products of political learning. The affect of policy actors' extremeness of beliefs participation in coalit ion building, and participation in facilitated multi stakeholder consensus based process es on political learning will be examined This study is based on a survey administered in 2011 and will employ the lens of the Advocacy Coalition Framework (ACF) to help examine some of the factors that promote and shape the political learning of policy actors involved in climate and energy policy debates in Colorado The results indicate that extreme beliefs are related to increased reinforcement of advocacy strategies, and that collaboration with indiv iduals with differing policy views is associated with more balance and changes, as opposed to reinforcement, of advocacy strategies

PAGE 128

! 114 INTRODUCTION A crucial reason to better understand the learning in policymaking is the need for sustainable policy to add ress threats such as climate change. Learning in the policy process has been argued as a normative goal for sustainability (Crona & Parker, 2012; Reed et al., 2010). Despite diversity in value laden beliefs, policy actors should, it is argued, come togethe r to exchange ideas; and in this exchange cognitive changes can take place and improved policy outcomes can arise (Henry, 2009; Murro & Jeffery, 2008). In this way, learning could be considered a necessary, if not sufficient, condition for sustainability. A step between individual cognitive learning and policy change is individual behavioral learning (in other words putting the new ideas into action), for instance a change r egarding political strategies. Policies are created through the strategic advocacy of multiple interact in g groups and coalitions of specialized individuals ( Kraft & Furlong, 2015; Birkland 2011 ). These specialized individuals often referred to as policy actors have invested time and resources gaining knowledge regarding specific pol icy issues as well as building political networks as part of their goal to affect the development of public policy through strategic advocacy ( Jenkins Smith et al., 2014 ). Focusing on these policy actors, this chapter examines the factors that shape indiv idual political learning regarding change or reinforcement of advocacy strategies. The concept of learning has long been a part of policy process theory. The policy process, in which problems are conceptualized and solutions are attempted and revised over time, was early on labeled as muddling through different variations of policy ideas and learning from this process (Lindblom, 1956). This conceptualization of learning may

PAGE 129

! 115 have been in reaction to early approaches to policymaking as systems theory (Easto n, 1953) or as a purely political system of stages of societal demands and policymakers reactions (Grin & Loeber 2007). In Heclo's (1974, p. 305) seminal study on political learning he suggested the role of this uncertainty and collective puzzlement as an alterative to political power in explaining the policy process. Similarly, Rose ( 1991 1993) conceived of "l esson drawing and learning by civil servants implementing new knowledge as a driver of policy Sabatier (1988, p. 133), influenced by both t hese and other top down and bottom up conceptualizations of learning in the policy process in his introduction of the Advocacy Coalition Framework (ACF) defined learning as continuing revisions of belief and behaviors related to achieving policy goals based on new experiences and information and highlighted learning as a integral part of the ACF's ability to explain the policy process. Over time multiple theoretical frameworks used to understand policy process have come to make learning a central compo nent explaining collective action and policy change (Heikkila & Gerlak, 2013). Thus, models and theories explaining individual learning now abound in the public policy literature. Building on Sabatier's definition, M ay (1992) distinguishe d between: 1) poli cy learning, relating to policy actors' social construction of policy problems and viability of specific implementation design ; and 2) political learning, related to policy actors becoming more sophisticated in their advocacy for policy problems and idea s This difference, and May's definition of political learning specifically, will inform the analysis described in this chapter. Previous analysis described in Chapter Three examined policy or cognitive learning ( belief change and reinforcement ) in this samp le of policy actors.

PAGE 130

! 116 This examination of learning as related to belief change is common in theories and applications of learning in the policy process (Heikkila & Gerlak, 2013; Birkland, 2011; Reed et al., 2010; May, 1992). In contrast, t he focus of this c hapter is on those factors that affect political or behavioral learning An example of political learning might be changing the target of advocacy strategies (e.g. a different committee, level or branch of government) (May, 1992). But changing advocacy strategies is not the only possible outcome of political learning. Learning is occurring in "In a world of scarce resources, those who do not learn are at a competitive disadvantage in realizing their goals" (Jenkins Smith & Sabatier, 1993 p. 44 ) In othe r words, political learning is strategic in nature. Policy actors may try out new media tactics attempting to call attention to problems or policy, but also might emphasize advocacy strategies that have worked. In other words, advocacy strategies may be ch ange d or reinforced. This chapter presents analysis aimed at examining factors that promote political learning, as well as different products of that learning, namely changes or reinforcement of policy actors' advocacy strategies. Better understanding the factors that affect political learning is essential to gaining a clearer and more complete picture of policy actors within the policy process, and leads to the r esearch question of this study: What factors affect policy actor political learning in a local and state level energy and climate policy subsystem? Studying advocacy strategies directly, as opposed to policy beliefs, allows for a more direct examination of the specific behaviors policy actors engage in to further their policy goals (Heikkila et al., 2014). In an effort to better understand factors shaping

PAGE 131

! 117 political learning, this study will use the ACF as a lens to examine policy actor politic al learning in the form of change or reinforcement in advocacy strategies in the climate and energy policy subsystem of Colorado. THEORY AND HYPOTHESES : POLITICAL LEARNING AND THE ADVOCACY COALITION FRAMEWORK To contextualize this study, some discussion o f the examination of learning across disciplines is warranted. Natural sciences such as neuro biology have been struggling to understand behavioral learning since its roots in the late 1800s (Milner, Squire & Kandel, 1998) Similarly, social scientists such as s ociologists have sp ent much time examining factors shaping learning regarding advocacy strategies and collective action as relate d to social movements and policy change ( Tilly & Tarrow, 2007). Despite decades of studies, cross disciplinary comparison s are complicated by the lack of a singular definition, concept ualization or operationalization of learning. That said, learning in policymaking is often cited as a normative goal and even as a potential requirement for sustainable political systems (Reed et al., 2010; Henry, 2009; Parson & Clark, 1995). It therefore seems crucial to delineate and understand different definitions of learning in policy process research before examining the factors that might shape policy actor political learning. The concept of learning has been and continues to be integral to many policy process theories, models, and frameworks including the Advocacy Coalition Framework, the Narrative Policy Framework and the Diffusion of Innovation models to name just a few ( Jenk ins Smith et al., 2014 ; McBeth, Jones & Shanahan, 2014; Berry & Berry, 2014; Heikkila & Gerlak, 2013 ; Birkland, 2011 ). But there is diversity in the way learning is

PAGE 132

! 118 conceptualized in the policy process literature. May (1992) offered a distinction between policy learning and political learning. In May's typology policy learning has two dimensions: social policy learning regarding the potential changes in ways policy problems and policy goals are social ly constructed, and instrumental policy learning relati ng to refined understanding of specific policy instruments or implementation designs. Previous analysis examined policy learning in the same sample of policy actors and this chapter will focus on political learning. May (1992) defined p olitical learning a s learning regarding political strategies, such as changes ( or reinforcement s) of policy actor advocacy strategies in advancing policy ideas. Contemporary policy process research is defined as the investigation of interactions between public policy and the pertinent actors, occurrences and contextual factors as well as policy outcomes (Weible, 2014b, p. 5) The study of these interactions can now be approached from a diverse set of theories, models and frameworks (Cairney & Heikkila, 2014). One example is t he Advocacy Coalition Framework (ACF), developed by Sabatier and Jenkins Smith in the late 1980s after decades of research on environmental problems as an attempt to link botto m up theories of the policy process with top down ideas of policy implementation (Sabatier 1988 1998). The ACF has become a leading framework to examine and exp lain the role of individuals and coalitions in the policy processes, learning in the policy process, and policy change ( Jenkins Smith et al., 2014; Birkland, 2011; Freeman, 2006). As with many theories and models in social science s the ACF builds on the m odel of the individual as boundedly rational possessing perceptual limits and filters including preexisting beliefs (Simon, 1985). Research under the ACF attempts to

PAGE 133

! 119 understand and explain the policy process by examining the different individuals and grou ps (coalitions) marshaling their available resources and employing advocacy strategies in order to influence policy outputs. The resulting policies and programs can be thought of as the translation of the beliefs of policy actors from one or more coalition s (Jenkins Smith et al., 2014). This stems from an assertion that individual beliefs are the driver of political behavior and pull like minded actors together in advocacy coalitions (Sabatier, 1988; Sabatier & Jenkins Smith, 1993) The ACF defines advocacy coalitions as group s of policy actors, (e.g. legislators, government employees private and non profit sector inte rest group leaders) who share policy beliefs and goals and collaborate in their attempts to influence the policy process ( Jenkins Smith et a l., 2014 ). The policy actors as individuals, as opposed to their advocacy coalitions, will be the u nit of analysis for this study and t he focus of this analysis is not policy learning in terms of belief change, but political learning. Policymaking occurs among these specialized policy actors working with in the policy subsystem according to the ACF (Jenkins Smith et al., 2014). A policy subsystem is the larger context of coalitions ( e.g. government agencies, private interest groups, and non profits ) and policy actors involved in a specific or specialized topic. A subsystem spans the substantive policy topic (in this case climate and energy policy) in a geographic area (in this case Colorado). The policy actors shape the policy landscape by establish ing policies through advocacy strategies and frame and narrative s and arguments in pursuing policy goals (Heikkila et al., 2014; Shanahan et al., 2013).

PAGE 134

! 120 Theoretical Emphasis The ACF can support multiple theoretical emphases such as: advocacy coalitions formation and changes over time, policy actor and coalition resources and advocacy strategies, policy change, and learning. The ACF provides numerous models of causal mechanisms and hypotheses regarding all of these theoretical emphases (for a full list of hypotheses see: Jenkins Smith et al., 2014) but learning in the policy process, and political learning in particular, is an im portant concept within the ACF. Not only does Sab atier (1988) employ the word learning in the title of the article introducing the ACF, but explicitly discusses examples of political learning such as coalitions learning to shift tactics toward challenging opponents causal arguments and changing strate gies to expand and mobilize groups. However, learning has been inconsistently conceptualized and measured across applications of the ACF O ne reason for this perhaps is that learning has been the least explored of the three major ACF theoretical emphases ( Jenkins Smith et al., 2014) The ACF explicitly argues that individuals and groups use strategies to advance policy goals and that those strategies may shift in some learning process (Sabatier & Jenkins Smith 1993). However, the majority of ACF arguments focus on how this learning may refine policy actors belief, such as their understanding of logical and causal relationships regarding specific policy issues or issues related to the problem attempt ing to be addressed by policy (Sabatier & Jenkins Smith, 1 993: pp. 42 43; Heikkila et al., 2014; Jenkins Smith et al., 2014). Far less attention has been paid to the advocacy tactics employed by policy actors, and the political learning processes and products, or results of learning, that occur in the policy proc ess (Shanahan et al., 2011; Weible et al., 2009).

PAGE 135

! 121 This chapter, in an attempt to contribute to the theory and measurement of models of individual learning, will build on the ACF to test new hypotheses related to policy actor political learning. This chapter specifically conceptualizes and defines behavioral reinforcement in addition to behavioral change, as evidence of political learning Reported behavioral, as opposed t o cognitive, change and reinforcement represent this study's concept of the resul ts (or products) of learning. While political behavioral change and behavioral reinforcement may both be included in a concept of learning in the ACF, this study argues they are opposite kinds of political learning. In other words, changes to and reinforce ment of policy actor advocacy strategies are both positive cases of learning occurring ( Goertz 2006). Since political learning is defined in this study as including change and reinforcement in advocacy strategy as well as different categories in between, then reports of neither change, nor reinforcement would conceptually be negative cases or "not learning" ( a lack of political behavioral change or reinforcement ) (Goertz, 2006) In other words, if policy actors report no change or reinforcement of advocacy strategies across categories of responses, this is considered non learning (the absence of behavioral/political learning). Heikkila and Gerlak (2013) point out that the notion of non learning is conceptually difficult in a recent meta analysis of learning i n public policy literature Given that t he purpose of this chapter is to examine the factors that promote political learning, as well as different resulting forms that learning might take, such as advocacy strategy change or reinforcement, and with focus o n clear conceptualization and measurement of learning, the lack of behavioral change or reinf orcement is defined as non learning.

PAGE 136

! 122 The ACF contends policy actors with more extreme policy beliefs may be more constrained in a learning process regarding changing their beliefs (Jenkins Smith et al., 2014). Previous analysis from this same sample of policy actors described in Chapter Three found that extreme views were associated with policy learning i n the form of belief reinforcement, as opposed to belief change. These findings support previous ACF applications regarding within coalition learning taking the form of policy belief reinforcement (Litfin, 2000; Meijerink, 2005) Given the importance of in dividual belief structures within the ACF regarding filtering and interpreting information, extreme beliefs may affect political learn ing as well as policy learning. It may be that having more extreme views regarding specific policy issues may make policy actors less flexible to change behaviors such as advocacy strategies. However, as political learning is strategic in nature, change or reinforcement of strategies is based on desire to attain policy goals (Jenkins Smith & Sabatier 1993; May 1992). Theref ore if policy actors with extreme views, and presumably more impetus to succeed in a competitive policy process, advocacy strategy reinforcement (of successful strategies) may be as likely to be associated with extreme belief as advocacy strategy change ( refinement of existing strategies ) In either case, extreme views are not likely to be associated with the lack of either behavioral change or reinforcement. In other words, extreme views may promote political learning. B uilding on th is argument, this chap ter will include results of analyses that t est the following hypothesis: Hypothes i s 1: Policy actors with extreme beliefs will be more likely to change or reinforce advocacy strategies than not change or reinforce advocacy strategies.

PAGE 137

! 123 It is postulated in the ACF that actors' participation in specific policy activities may also affect learning ( Sabatier & Jenkins Smith 1993 1999). For example when p olicy actors engage in more interaction between individuals with dissimilar policy views and goals (sometimes called cross coalition activities ) this could lead to more cross coalition learning. The ACF argues there are conditions where this cross coalition learning is more likely such as: when issues a re more tractable and technical in nature (as opposed to socio cultural), when conflict is at intermediate levels, and when f orums are present that are respected enough to facilitate policy actors from different coalitions to participate in ways dominated by professional norms (Jenkins Smith et al., 2014 p. 200 ). Some previous ACF applications have shown evidence supporting these claims (Meijerink, 2005; Elliot & Schlaepfer, 2001; Olson et al., 1999). However, other applications (Litfin, 2000; Munro, 1993) have found these conditions do not always lead to cross coalition learning. Most of these applications have examined learn ing as belief change or policy learning, with little attention paid to explicitly examining policy actors' political learning. Applic ations that have examined changes in advocacy strategies (Albright, 2011; Nohrstedt, 2011) have done so at the coalition level, as opposed to the individual policy actor level, and have focused on the connections between changes in strategies and major eve nts in the subsystem, as opposed to attributes of policy actors (such as beliefs and policy activities) as factors that might shape political learning. Participation in coalition building would seem to be a critical policy activity to examine. ACF theories explicitly state the assumption that members of advocacy coalitions learn lessons and improve advocacy to meet policy goals accordingly (Jenkins Smith & Sabatier, 1993; Sabatier,

PAGE 138

! 124 1988 ). In fact May (1992, p. 340) explicitly states, " political learning t akes place within coalitions Likewise, a s mentioned, participation in forums such as negotiated multi stakeholder consensus based processes is argued by the ACF to expose policy actors to differing views that potentially help build shared (across coalitions) understanding of problems and alter advocacy strategies accordingly in order to achieve policy goals (Jenkins Smith et al., 2014). Therefore b oth participation in coalition building and participation in negotiated multi stakeholder consensus based processes are likely to be associated with political learning. Hypothesis 2: Policy actors that participate in coalition building will be more likely to change or reinforce advocacy strategies t han policy actors that do not participate in coalition b uilding. Hypothesis 3: Policy actors that participate in facilitated multi stakeholder consensus based processes will be more likely to change or reinforce advocacy strategies than policy actors that do not participate in negotiated multi stakeholder consensus based processes The definition in this study of political learning includes both advocacy strategy change as well as reinforcement Given this, another level of theorizing and analysis is permitted here Reinforcemen t of advocacy strategies is different than change of advocacy strategies. A reinforced advocacy strategy could be a strategy that is more firmly held, more supported from successful experience. However, reinforcement of strategies may also occur when policy actors are not exposed to new ideas for advocacy strategies, or have their understanding of policy and problems change (e.g. primary causes of climate ch ange, or pathways to improve the political palatability of a cap and

PAGE 139

! 125 trade system to address climate change), thus stimulating a refinement (change) of strategies (Sabatier, 1988). In this way, coalition building can provide policy actors with more raw mat erial of experience and interaction with more policy actors with similar goals, l eading to changes (alterations) in advocacy strategies. In other words, this within coalition learning would presumably lead to more change in advocacy strategies than reinfor cement of strategies. Conversely, policy actors that do not participate in coalition building simply have less diversity of experience and new information and are likely to be exposed to fewer new i deas for arguments and tactics, and thus more likely to e xperience reinfo rcement of advocacy strategies. A similar argument could be made for policy actor participation in multi stakeholder consensus based processe s In these forums, cross coalition learning is possible when policy actors with different policy g oals discuss policy relevant facts, make causal arguments linking problems and policies, and espouse strategic advocacy arguments and tactics (Jenkins Smith & Sabatier, 1993; Sabatier 1988). Therefore these facilitated multi stakeholder consensus based processe s would offer policy actors the opportunity to hear the arguments and tactics of political foes and would provide a wealth of experience and diversity of information leading to refinement and revision (change) to advocacy tactics. Policy actors that do not participate in multi stakeholder consensus based processe s would have less exposure to this diversity of argument and strategies presented, and would be more likely to experience policy learning as advocacy strategy reinforcement. Building on t his argument, this chapter will present analysis testing the following two hypotheses with the goal of extending actor based models of political learning in the ACF.

PAGE 140

! 126 Hypothesis 4: Policy actors that participate in coalition building will be more likely to report advocacy strategy change than advocacy strategy reinforcement. Hypothesis 5: Policy actors that participate in facilitated multi stakeholder consensus based processes will be more likely to report advocacy strategy change than advocacy strategy reinforcement. CLIMATE CHANGE POLICY The Intergovernmental Panel on Climate Change and broader scientific communities have concluded unambiguously that the earth is experiencing catastrophic anthropogenic climate change, sometimes called climate destabilization (IPCC, 2014; Giddons, 2011; Stern, 2007). This will have dramatic and potentially irreversibly destructive impacts on human populations (IPC C, 2014 2013). In the face of this threat, the U.S. federal government thus far refused to ratify the Kyoto Protocol, the international agreement to address climate disruptio n, or create any meaningful and comprehensive national climate or energy policy a imed at addressing the issue (Vig & Kraft 2015; Layzer, 2006). 14 One reason for this may be the conflicting goals and contrasting political strategies of policy actors in favor of climate policy and those with political and economic interest in resisting c hanges to, for instance, national energy policies (Vig & Kraft, 2015; Giddons, 2011). Th is has stimulated a large number of subnational government entities to creat e climate and energy policies of their own. L ocal governments and states have now created an d implemented a vast array of programs and !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! !!!!!!!!!!!!!!! 14 This might change by December 2015 as President Obama has set goals and is a ttempting to have the U.S. be a part of a new international climate agreement, but details and consequences of these efforts are, at this point, far from certain. President Obama has proposed some national policies aimed at changes to energy generation to reduce greenhouse gas emissions, but the results of implementation of these rules are also uncertain.

PAGE 141

! 127 legislation T he landscape of subnational climate and energy policy has thus become a natural laboratory for political scientists and policy scholars to try to test various hypotheses regarding policy actors and the policy process. Subnational Climate Policy Landscape In the absence of federal action a number of subnational climate policy programs have been created such as the U S Mayors Climate Prote ction Agreement (USMPCA) which began in 2005 and is an attempt to advance the goals of the Kyoto Protocol through local government leadership and action ( US Conference of Mayors, 2015 ). The USMPCA now has more than 1,000 signatories, including Denver, Colora do (US Conference of Mayors, 2015 ). Signatory c ities agree to reduce community wide greenhouse gas (GHG) emissions to at least 7 percent below 1990 by 2012 In an attempt to achieve these emission reduction targets cities have over the years promulgated litany of climate related polices and programs 15 Subnational climate policy has also been promulgated at the state level. Over 30 U.S. states, including Colorado, have also created and published a specific climate action plan (CAP) of some sort ( EPA, 201 5 ). A CAP typically states the state or local government's GHG emission reduction goals and explicates the bundle of policies that w ill be implemented to reach those goals Like state CAPs, hundreds of cities and counties (many USMPCA signatories) in the U.S. have also created CAPs that outline the policies and programs that will be employed by these stats and local gove rnments to reac h stated goals. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! !!!!!!!!!!!!!!! 15 For a review of the results of these trends see Krause, 2011 and Wood et al ., 2014.

PAGE 142

! 128 Denver and CO Climate P olicy Landscape Denver has a history of climate policy going back to the mid 1990s as an e arly member of the International Council of Local Environmental Initiatives (now know n simply as ICLEI) ( for a review see: B ulkeley & Betsill, 2003). Then Mayor John Hickenlooper created Greenprint Denver in 2005. It was a new department within the City and County of Denver with the goal of further integrating environmental impacts along with economic and social analysis into c ity programs and policies (City and County of Denver 2006 ). The creation of this local program was laudable and the stated goals and current progress thus far are typical of a USMCPA signatory for that time period ( US Conference of Mayors, 2015 ). Denver i s also typical compared to other medium to large U S cities that are signatories to the USMCPA in terms of its population density Around the same time changes were happening at the City and County level, the state of Colorado was also taking steps to cha nge climate and energy policy. T hen Governor Bill Ritter launched an initiative to address climate change statewide in 2007 which resulted in the creation of the Colorado Clim ate Action Plan The Colorado CAP calls for a reduction of state emission s of gr eenhouse gases by 20 percent below 2005 levels by 2020 and a longer term goal of 80 percent below 2005 levels by 2050 ( Colorado Climate Action Plan: A Strategy to Address Global Warming, 2007 ) Similar to Denver compared to its peers, in terms of scale and scope of policies and emission reduction targets, the Colorado CAP is typical of state CAPs (EPA, 2015; Ramseur, 2007). For these reasons Colorado represents a typical case compared to similar states (Gerring, 2007).

PAGE 143

! 129 Mayor Hickenlooper in now the Governor of Colorado, overseeing the implementation of the state CAP. For the purposes of this study the policy actors from across Colorado, including those from the City and County of Denver, will be treated together as subnational actors. One reason this is justified is th e overlap in city and state political actors over the years between 2005 and now. Also, Denver is the largest city in Colorado, the state's capital, and influential in state level po licy development s in climate and energy issues. Lastly, as is often the case across many issues, the state and cit y governments have an integrated approach to energy and climate policy Activities and Advocacy Strategies of Colorado Climate and Energy Poli cy Actors A brief description of the Colorado climate and energy policy actor demographics is warranted. Elgin and Weible (2013) conducted related but different analys e s on the same sample of policy actors presented in this chapter. Their work using cluster analysis to measure the relative proportions of the same six policy beliefs among the same policy actors used in this study identified two distinct advocacy coalition s in this sample. A proclimate coalition consisting of 205 of the 260 policy actors (79 percent ) that believed : climate change will have severe impacts, humans are the cause of climate change the government should intervene in the marketplace to address climate change and were supportive of multiple policies to address climate change. The second coalition, the anticlimate coalition contained 55 of the 260 policy actors (21 percent ) with significantly different opposing views. See Elgin and Weible (201 3) for a full summary of methodology used to determine advocacy coalition membership but of note is that the proclimate coalition contained significantly more government employees, as a percentage of the coalition members, than the anticlimate coalition, which was dominated by

PAGE 144

! 130 members of the business community. Secondly, proclimate coalition members identified themselves as significantly more liberal on fiscal and social issues compared to policy actor s in the anticlimate coalition. In terms of specific ac tivities and advocacy strategies employed by these policy actors, Elgin and Weible (2011 p p 128 129 ) found the two coalitions had very similar levels of activity regarding: appraising policy options, conducting relevant research, consulting with the public, implementation of policies and programs, informing officials, and participation in multi stakeholder consensus based processes. S ignificant differences in activities and strategies were observed across organizational affiliation. For inst ance, academic researchers were more likely to conduct relevant research, and government employees were more likely to implement polices and programs. In the conclusion of their analysis and Weible (2013 p. 130) state, despite different policy beliefs, th e two coalitions were relatively similar in many regards, including the use of similar activities and political strategies. Despite the fact that both coalitions were judged to have high individual and organization capacity to engage in debates, given the relative size, the proclimate coalition was observed to be in a favorable position. Given the hypothesized relationship between participation in coalition building and political learning, and participation in multi stakeholder consensus based processes and political learning, these will be the political activities examined more closely in the analysis below. DATA AND METHODS Research Design This chapter presents findings from the last quantitative analysis of a larger study of learning in the climate and energy policy subsystem of Colorado. Data comes from an

PAGE 145

! 131 original cross sectional electronic survey of 260 policy actors and the ACF will be used as a theoretical foundation to investigate factors shaping political learning regarding advocacy strategy. Data Collection C onsistent with the literature, policy actors of the climate and energy policy subsystem in Colorado were the target population for the study 's survey Policy actors (sometimes referred to as policy elites) are defined as individuals with knowledge of and involvement in the policy subsystem The individuals were employed by government agencies, non profit organizations, and private companies involved in climate policy in Colorado In generatin g a list of study participants as the target population t hree major strategies were used. To start, individuals who attended the roundtable sessions preceding the creation of state and local government policies and those who attended relevant public input sessions d uring the process were targ eted for inclusion. Following that, a snowball sample was generated from the lists and with input from an informal advisory council T he survey was sent to 793 policy actors involved in climate and energy issues in Colorado ; 260 individuals returned fully completed surveys (response rate of 33 percent ) Data Analysis The survey data w ere analyzed using STATA 13 software. D escriptive statistics of all variables were calculated and are presented below A multinomial logistic regression model was used, as the dependent variable is a non ordinal scale to test the hypotheses related to extremeness of policy beliefs, participation in coalition building, and participation in multi stakeholder consensus based p rocesses on individual policy

PAGE 146

! 132 actor political learning. The relevant assumptions regarding the stati stical model, n ormality, skewedness, and m ulticollinearity were examined. Operational Measures To measure policy beliefs and extremeness of belief, respond ents were asked to indicate their relative agreement or disagreement on a 5 point Likert scale with the following six statements: 1) The severity of predicted impacts on society from climate change are vastly overstated ; 2) Human behavior is the princip al cause of climate change ; 3) Decisions about energy and its effect on climate are best left to the economic market, and not to the government ; 4) An energy and/or carbon tax is required to combat climate change ; 5) A cap and trade system of permits for the emission of greenhouse gases is required to combat climate change ; and 6) Government policies to promote renewable energy generation are required to combat climate change Policy activities were measured by asking respondents to report Yes or N o in the past year if they participated in : coalition building (e.g. networking, information sharing) and multi stakeholder consensus based process To measure the dependent variable, political lear ning regarding advocacy strategies respondents were asked: "To what extent have your strategies changed or been reinforced regarding the way you advocate for climate related issues and/or energy policy?" The six possible responses provided were as follows: 1) Only been reinforced, 2) Mostly reinforced with little changes, 3) Balance of reinforcement and changes, 4) Mostly changed with little reinforcement, 5) Only been changed, and 6) Neither changed nor been reinforced

PAGE 147

! 133 RESULTS AND ANALYSIS Descriptive Statistics Descriptive statistics in cluding e ach of the independent variables and dependen t variable are presented below. Extreme Beliefs The policy actors surveyed were asked to respond with their relative agreement to six policy belief statements on a 5 point Likert scale ( s trongly agree, s omewhat agree, n either agree nor disagree, s omewhat disagree, and s trongly disagree ). An Extreme Belief score was created by coding each answer in the following way: s trongly agree and s trongly disagree = 2 s Somewhat agree and s omewhat disagree = 1 and n either agree nor disagree = 0 These extreme b elief scores were then summed for all six policy belief questions to create a n aggregate Extremeness of Policy Beliefs variable for each of the respondents. The minimum Extremeness of Policy Beliefs score was 1 while the maximum score was 12 ( also the maximum possible based on coding technique ). The mean Extremeness of Policy Beliefs was 8.26 with a standard deviation of 2.34 16 As F igure 4.1 below shows the distribution of extreme beliefs in the sampled popu lation is skewed toward above the mean (more extreme). !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! !!!!!!!!!!!!!!! 16 This range comports with the individual extreme policy belief questions employed in the analysis presented in Chapter Three. Four of the policy belief questions had a mean extreme score of 1.45 1.48, and two of the policy belief questions had a mean extre me score of 1.02 and 1.23. While the aggregate extreme belief score skews toward the extreme, there is adequate diversity in answers provided to test the stated hypotheses.

PAGE 148

! 134 Figure 4.1 Policy Actor Extremeness of Beliefs Policy Activities : Coalition Building and Multi stakeholder C onsensus based P rocesses One hundred sixty five ( 165 ) of 260 ( 63 percent ) of policy actors answered "Y es th ey had participated in coalition building (networking and information sharing e.g.) in the past year. One hundred forty six ( 146 ) of 260 (56 percent ) indicated "Yes they had participated in facilitated multi stakeh older consensus based processes Dependent Variable: Political Learnin g Regarding Advocacy Strategies Respondents were asked: "To what extent have your strategies changed or been reinforced regarding the way you advocate for climate related issues and/or e nergy policy?" Potential responses were: Only been reinforced, Mostly reinforced with little changes, Balance of reinforcement and changes, Mostly chan ged with little reinforcement, Only been changed, and Neither changed nor been reinforced As can been se en in Figure 4.4 below, while Balance of R einforcement and C hanges was the modal response (76:30 percent ), reinforcement was more common than change. Neither C h anged nor B een R einforced (non learning) was also more common tha n change in advocacy strategy. '! '! @! &! ##! #"! =#! =&! =%! $A! ==! '=! "! '"! #"! ="! $"! >"! %"! '! #! =! $! >! %! @! &! A! '"! ''! '#! !"#$%&'()'*+,-.-,"/01' CR6&%#%'P%0-%)'28(&%')(&'/00'2-R'7(0-83'P%0-%)1' 7(0-83'<86(&'CR6&%#%+%11'()'P%0-%)1'

PAGE 149

! 135 Figure 4.2 Learning Regarding Advocacy Strategies Testing Hypothese s 1 5 The statistical method used to test the five hypotheses is explicated below. Multinomial Logistic Regressions The dependent variable for this analysis is political learning regarding advocacy strategies conceptualized as a set of categories of responses involving variations of change and reinforcement of strategies. This learni ng concept was operationalized as a set of categorical responses included a 5 point scale between a dvocacy change and advocacy reinforcement and a sixth possible response of neither changed nor been reinforced This sixth possible response means the dependent variable cannot be treated as ordinal scale. This is similar to including "I do not have an op inion" as a sixth option to a 5 point Likert agree/disagree question, which would invalidate the ordinal outcome models. In these cases the multinomial logit model (MNLM) also known as multinomial logistic regression (MNLR) is the most frequently use d mo del (Long & Freese, 2014). "! '"! #"! ="! $"! >"! %"! @"! &"! b60+!L))6! B)-6S2*:)3! 12740+! B)-6S2*:)3![-4C! ,-440)!5C/6T)7! L/0/6:)!2S! B)-6S2*:)F)64! /63!5C/6T)7! 12740+!5C/6T)3! [-4C!,-440)! B)-6S2*:)F)64! b60+!L))6! 5C/6T)3! H)-4C)*! 5C/6T)3!62*! L))6! B)-6S2*:)3! !"#$%&'()'*+,-.-,"/01' Q%/&+-+:'4%:/&,-+:'<,.(8/83'26&/6%:-%1'

PAGE 150

! 136 A multinomial logistic regression analysis was used to examine how the three independent variables affected the probability of different categories of learning responses being reported. In other words, how the independent variabl es shaped different products of learning. The MNLM is similar to aggregating a set of binary logits among all pair comparisons of alternative levels of the dependent variable, in this case the six levels of learning (Long & Freese, 2014; Long, 1997). For t his MNLR the base outcome was set to "belief reinforcement only" in all analyses. The raw outputs of MNLR are complicated and difficult to interpret intuitively but are presented in complete form in Appendix B of this dissertation. The most common and intu itive means of displaying the MNLM outputs a re those that help understand how marginal changes in the independent variables are associated with changes to predicted probabilities in the level of the dependent variable (Long & Freese, 2014). Marginal Change Plot The marginal change is defined as the average percent change (the probability of change) across the dependent variable associated with changes to each independent variable, holding all other i ndependent variables constant. A marginal change plot was calculated after the MNLR was run. In the marginal change plot below the average discrete change in the dependent variable based on one standard deviation increase in each independent variable (holding all others constant) are plotted together. Each outco me or level of learning is represented by a single letter: R = Reinforcement Only, r = Mostly reinforcement, B = Balance, c = Mostly Changed, C = Changed only, and N = No Change or Reinforcement.

PAGE 151

! 137 These plots represent changes in predicted probability. The resulting changes in the dependent variable based on the effec ts of the independent variables, are represented by shifts to the right or to the left (indicating more likely or less likely respectively ) from the center line of the graph of each outcome of learning represented by a letter Asterisks (*) indicate a significant relationship ( p < 0.05) between a change in the range of an independent variable and a change in the predicted probability between two categories of the dependent variable. In other w ords, if the probability of one outcome of learning is significantly affected either more likely or less likely ( indicated by placement to the right or left of the centerline) it will appear with an (*) indicating that change in probability is significant ly outside the predicted range. Measur e s of fit for the learning model are presented below with the Pseudo R 2 specifically McFadden's R 2 as it is the most common presented in a MNLR (Long & Freese 2014) 17 Collinearity diagnostics were performed for the MNLR and the tolerance and variance inflation factors (VIFs) indicated no multicolinearity issues !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! !!!!!!!!!!!!!!! 17 B ecause this statistic does not mean exactly what R 2 means in OLS regression (the proportion of variance in the dependent variable predicted by the independent variables), it should be interpreted with some caution.

PAGE 152

! 138 Figure 4.3 Changes in Predicted Probabilities in Learning Regarding Advocacy Strategy for Climate and Energy Policy Issues R = Reinforcement Only, r = Mostly reinforcement, B = Balance, c = Mostly Changed, C = Changed only, N = No Change or Reinforcement Pseudo R 2 = 0. 05 and the number of observations = 2 51 (*) indicates a significant relationship ( p < 0.05) btw a change in the range of an IV and change in predicted probability btw two categories of the DV Effects of Extreme Beliefs Policy actors' extremeness of beliefs was significantly associated with a greater likelihood of political learning in the form of reinforcement only of advocacy strategies. Extreme beliefs were also negatively associated with neither ch ange nor reinforcement or non learning regarding advocacy strategies. In other words extreme beliefs made political learning significantly more likely (nonlearning is less likely) and that learning was more likely to be shaped as reinforcement of advocacy strategies. Specifically, one R* r B c C N* R r B c C N R r B c C N Extremeness of Policy Beliefs Participation in Coalition Building Facilitated Participation in Multi-Stakeholder Consensus-Based Processes .15 .10 .05 -.05 -.10 -.15 Marginal Effect on Outcome Probability

PAGE 153

! 139 standard deviation increase in extreme beliefs (more extreme beliefs) resulted in policy actors being 12 percent more likely to report reinforcement only in their advocacy strategies and 8 percent less likely to report neither change nor reinforcement of advocacy strategies. These changes in predicted probabilities were both statistically significant ( p < 0.05). These results support H ypothesis 1: P ol icy actors with extreme views were significantly more likely to report political learning than to report political nonlearning. Going beyond the hypothesis, the results indicate that political learning associated with extreme beliefs was significantly more likely to be reported as reinforcement of advocacy behavior In other words, extreme policy beliefs were associated with political learning and were associated with political learning in the form of advocacy strategy reinforcement specifically. Effect of Participation in Coalition Building The results indicate no support for H ypothes i s 2: P olicy actor participation in coalition building was not associated with any change in probability of either political learning or not political learning. Ther e is also no support in these results for H ypothesis 4: P olicy actor s that participate in coalition building were not more likely to change, as opposed to reinforce their advocacy strategies. Changes in t his independent variable led to no significant chan ges in the predicted probability of a ny report ed outcome s of political learning Effect of Participation in Facilitated Multi Stakeholder Consensus Based Processes The results show no support for H ypothes i s 3: P olicy actor participation in facilitated multi stakeholder consensus based processes was not associated with any

PAGE 154

! 140 change in probability of policy actors reporting advocacy strategy change or reinforcement (political earning) versus no advocacy change or reinforcement (not political l earning). The re is also no support in these results for H ypothesis 5: P articipation in facilitated multi stakeholder consensus based processes was not associated with policy actors reporting change in advocacy strategies with greater likelihood than reinforcement of th eir advocacy strategies. This independent variable di d not le a d to any significant changes in the predicted probability of any of the six outcomes of political learning DISCUSSION The purpose of the analysis presented in this chapter was to better understand some of the factors that shape individual political learning regarding advocacy strategies in the policy process. Different variables were hypothesized to shape policy actor political learning differently, in other words, l ead to different forms of political learning. The results indicate extreme policy beliefs are associated with a greater likelihood of political learning as hypothesized, but also are associated with a specific product of political learning, behavioral rein forcement. There was strong support for H ypothesis 1. The extremeness of policy actors' beliefs was positively associated with political learning (nonlearning was less likely) This finding supports the longstanding ACF claim that policy actors will learn regarding advocacy strategies, resulting in changes to unsuccessful and/or reinforcement of successful strategies (Sabatier, 1988). However, s ince nonlearning was included as a possible outcome however the results here indicate an additional claim could be made. Policy actors with extreme beliefs are more likely to report political learning than not politically learning. In other words, in this subsystem, extreme beliefs promote political

PAGE 155

! 141 learning in policy actors. Perhaps policy actors with more extreme beliefs have a greater desire to see their policy goals attained and thus are more likely to either change or reinforce their advocacy stra tegies than not. Given the way learning was operationalized in this study there is one additional finding regarding the products of learning regarding advocacy behavior (as shaped by extreme beliefs). Namely, that extremeness of beliefs was also positively associated with reports of reinforcement only of advocacy strategies. In other words, having extreme beliefs made political learning more likely in policy actors b ut t he increased learning associated with more extreme beliefs was much more likely to be reinforcement of advocacy behavior. These findi ng corroborate the previous analysis presented in Chapter Three that examined the role of policy learning in this sample of policy actors Namely, that extremeness of beliefs had a similar positive relationship regarding political learning as reinforcement of advocacy strategies being more common than change as it did to policy belief reinforcement being more common than change The logic behind why this might be true regarding political learning is uncertain. If, as show n above, policy actors with extreme beliefs are more likely to politically learn (either change or reinforce their advocacy strategies), there are at least two reasons why advocacy strategy reinforcement is more common than change. First, it is possible that those with extreme beliefs are ac tually more successful policy advocates, and thus have more winning strategies that need only be reinforced. While this is possible, this study was not set up to capture that causal mechanism. A second possible causal mechanism for an association between extreme beliefs and advocacy strategy reinforcement is that similar to selection bias, where individual selective filters may be

PAGE 156

! 142 biased to new information and experiences (Lord et al., 1974; Simon, 1985) policy actors with extreme beliefs are merely convinced their advocacy strategies are winning. This may mean policy actors with extreme beliefs are rejecting evidence their advocacy strategies should be changed, thus resulting in increased advocacy str ategy reinforcement a n equally interesting causal logic, but also untestable in this study. The results provided no s upport for H ypotheses 2 3, 4, or 5. Neither participation in coalition building or participation in facilitated multi stakeholder consensu s based processes had any effect on promoting political learning, nor any effect on shaping political learning in the form of advocacy strategy change or reinforcement. This would seem to contradict the existing ACF claims of coalition building leading tow ard within coalition learning and cross coalition learning from activities such as forums bring ing diverse policy actors together Established ACF arguments state these activities would stimulate learning including political learning r egarding advocac y strategies ( Jenkins Smith et al., 2014; Jenkins Smith & Sabatier 1993 ; Sabatier, 1988 ). Why these policy activities would have no effect on po litical learning is not clear. One possible reason neither of these policy activities was associated with promoting or shaping politic al learning, is that individual policy actors in this subsystem do not conform to the assumption of the ACF. In other words, perhaps the sampled policy actors are guided in their decisions to change or reinforce advocacy strateg ies only by their internal (extreme) policy beliefs, and not by learning lessons from any interactions with individuals with either similar or dissimilar policy goals. If this is true that would indicate the respective anticlimate and proclimate coalition s found by Elgin and Weible (2013 ) exist in the sense that each is comprised of policy actors that share beliefs (and

PAGE 157

! 143 that those two groups have different beliefs), but that coalition building that was occurring (and a majority of actors reported they did engage in this behavior), is potentially not sufficient to affect political learning. The same could be said of facilitated consensus based processes, even though just over a majority of policy actors reported they participated in this activity, it too wa s perhaps not sufficient to affect political learning. These findings should be interpreted in light of study limitations. First, this is a study of one policy subsystem in one state and thus generalizing findings t o analogous policy processes in other sub national entities should be done carefully (Gerring, 2007). However, this policy subsystem represents a typical case in that Denver (the most populous city and capital of Colorado) is a member of the US Conference of Mayors Climate Agreement and both Denv er and the State of Colorado have state d typical climate goals and made typical progress compared to other cities and states ( EPA, 2015 ) This would indicate that the Colorado policy subsystem investigated here is representative of other similar policy subsystem s thus threats to external validity and case selection effect are mitigated to some extent Also, single case (in this case subsystem) studies are explicitly designed to examine specific independent variables as factors a ffecting specific depende nt variable s as opposed to broad comparative observations, and as such a higher degree of internal validity in the relationships between variables in this case is gained ( Gerring 2007 ; George & Bennett 2005) As to how generalizable the findings of this study might be to policy subsystems different from a local government climate and energy subsystem, some factors are worth considering. First, subsystems with as prominent a role of science and technical

PAGE 158

! 144 information as climate and energy debates might be more comparable. Similarly, the results of Chapter Two, as well as Elign and Weible's (2013) analysis showed that this sample of policy actors was relatively homogenous in terms of education, race/ethnicity, analytic training, policy activities, and advoc acy strategies. In other words, while significant differences were shown between policy beliefs across two coalitions, the policy actors in th is sample share many attributes. A policy subsystem with similar characteristics might be more comparable to findi ngs of this study Another potential limitation in these data is that there may be greater diversity in policy actors' attributes and behavioral learning than was captured in the snowball sample for this study. For example, Denver is the capital city (higher population and greater connectivity to state level government issues) and the total number of government actors in the sample may be skewed toward the City and County of Denver employees, and less inclusive of smaller and/or more rural communities in Colorado. Since this study measured all levels of government agencies (e.g. state and city) together, future studies may be required to further dissect potential differences in attributes such as extremeness of belief and collaboration with those with dissimilar views and behavioral learning across different sub populations of government employees. A similar argument could also be made for the business sector, academic/researchers, and non profit employees. Lastly like all self reported opinion based research, the results of this survey are potentially biased. For example, reports of learning are self reported and thus potentially li mited by respondents' memory understanding of the question, or honesty.

PAGE 159

! 145 CONCLUSION The goal of this chapter was to examine factors that may shape individual policy actor s' political learning in the policy process. To that end, this chapter tested new hypotheses building on the foundation of the ACF. This is one of the first ACF applications to specifically ask policy a ctors to report political learning as advocacy st rategy change and reinforcement, and explicitly ask about the source of that political learning. Previous ACF applications examining changes in advocacy strategies have been in the context of explaining poli cy change or major events in the subsystem (Nohrstedt, 2011; Albright 2011; Hirsh et al., 2010; Nohrstedt, 2005). Ideally this study stimulates similar applications and adds to the undeveloped theorizing of the ACF regarding individual political learning regarding advocacy strategies. Given the way political learning was conceptualized and operationalized in this study, the results paint an interesting picture regarding the way extreme beliefs shape political learning. Since this study specifically asked about the relative incidence of no change or reinforcement in advocacy strategies (not learning) this study concludes extreme beliefs promote political learning, learning in the form of reinforcement of political behaviors. In other words, policy actors w ho reported more extreme views were significantly more likely to report advocacy strategy reinforcement only and less likely to report neither change nor reinforcement in advocacy strategy Political learning includes reinforcement of strategy behavior, s o having extreme views does not impede policy actor learning. This is the major finding in the analysis summarized in this chapter. Within coalition activities and cross coalition activities had no effect on political learning in this study. Collectively the findings here may improve upon existing ACF

PAGE 160

! 146 models of individual learning, as well as other policy process theories and frameworks (Heikkila & Gerlak, 2013). For instance, the Narrative Policy Framework builds on the concept of policy actors' belief sy stems and emphasizes the role of narrative strategies used to influence coalition and public opinion (Mc B eth, Jones & Shanahan, 2014) Perhaps the role of extreme beliefs in promoting political learning, which was shown to lead to changes in advocacy strat egy here, are reasons narratives might change In subsystems with policy actors with extreme beliefs, perhaps narratives will shift faster or will shift more dramatically in narrative. If so, shifting narratives may be a result of political learning and p oint toward another way this concept could be measured in future research. Previous stu dies have shown a connection between this narrative change and policy change (Shanahan, Jones & McBeth 2011; Jeon & Haider Markel 2001). Conversely, the results of t hi s chapter also indicate it could be interesting to examine the effect of stories and narratives being told on those individuals identified as having extreme beliefs compared to those with moderate beliefs Relatedly, a recent application attempting to int egrate individual level theories of learning with policy innovation and diffusion models recently showed ideological predispos itions affected lessons learned from various political approaches to policy problems (Butler et al., 2015). Perhaps the role of ex treme belief promoting learning and shaping learning in the form of advocacy strategy reinforcement seen in results described in this chapter would be useful to scholars working to improve the integration of individual level learning theories with related policy models operating at different units of analysis, such as diffusion and innovation of ideas between governments or sta tes (Berry & Berry, 2014).

PAGE 161

! 147 One lesson learned from this analysis is that it would have been beneficial to link political learning conceptually and operationally to policy actors' satisfaction and/or dissatisfaction with different policy outcomes. In other words, it would have been p referable (in hindsight) to more directly link policy actor s' political learning with the perception of their advocacy strategy effectiveness. This may have allowed the precise causal mechanisms behind the effect of extreme beliefs to be unpacked further. This cou ld be a part of future studies. Relatedly, attributes (institutional arrangements) of forums such as facilitated consensus based processes may vary. For instance forums may vary in degree of openness of participation and extent of norms of profess ional conduct. Existing ACF theories suggest attributes such as these may determine the extent of cross coalition learning, including political learning that may occur (Jenkins Smith et al., 2014). A lesson learned from this analysis is that additional que stions could have determined the perceived difference in these forums, which may have added insight to the causal mechanisms involved here. Again, future studies could add q uestions that would probe here. The results presented in this chapter clarify some of the factors that shape individual political learning regarding advocacy strategies in the policy process. Extreme policy beliefs are associated with a greater likelihood of political learning, but also are associated with a specific product of political learning, behavioral reinforcement. A t every step in the analysis presented in this chapter, every effort was made to emphasize clear conceptualization and measurement of different products of political learning and factors that might promote political le arning. Hopefully these findings can contribute to efforts to improve models and hypotheses of individual le arning in the ACF and beyond.

PAGE 162

! 148 CHAPTER V CONCLUSION DISSERTATION SUMMARY The main goal s of this dissertation were to examine factors that shape policy and political learning in policy actors and to contribute to the public policy literature working to understand the complexity of the policy process and compare different theories trying to explain it. More specifically, the dissertation aimed to study and bet ter understand relationships between cognitive filters and the interactions between policy actors and their policymaking contexts to policy actor learning To that end, the role of existing beliefs and policy activities in shaping policy and political learning were measured. Change and reinforcement in beliefs and behaviors were included in the analysis of learning with an effort to improve upon our understanding of the concept of learning and the growing know ledge of the role of learning in the policy process. With the objective of using shared vocabulary and established definition s of concepts, the lens of the Advocacy Coalition Framework (ACF) was used to examine some of the factors that shape policy learning and political learning. While connections were made to other policy process theories throughout the dissertation, the consistent use of the ACF attempted to clarify assumptions such as the model of the policy actor and the relationships between key concepts examined. Using a cross sectional case study, t he unit of analysis is individual policy actors involved in climate and en ergy policy debates in Colorado and a sample of this population was surveyed i n the spring of 2011

PAGE 163

! 149 CHAPTER TWO OBJECTIVES AND SUMMARY FINDINGS The first major objective of th e analysis presented in this chapter was to understand and describe attributes of the policy actors involved in climate and energy policy debates in Colorado, including their beliefs across a number of relevant policy questions The results show this sample of policy actors vary throughout many attributes including organizational affiliation and a number of policy activities, as well as across the poli cy beliefs examined. The second major objective of this study was to determine the extent of policy and political learning, as defined as a range between belief change and belief reinforcement and changes and reinforcements in advocacy respectively in the sampled population of policy actors. In general regarding policy learning, across all six major policy beliefs, reports of b elief reinforcement were much more common than reports of belief change. Similarly, in terms of political learning, reports of rein forcement of advocacy strategies were more common than reports of change in strategy. Reports of neither change nor reinforcement of beliefs and advocacy strategies was also more common th at belief or behavior change. CHAPTER THREE RESEARCH QUESTION, HYPOTHESES, AND SUMMARY FINDINGS The objective of the analysis presented in Chapter Three was to examine f actors shaping policy learning. Research Question: What factors affect individual policy actor belief change and reinforcement in a local and state le vel energy and climate policy subsystem? Table 5.1 below summarizes the Chapter Three hypotheses and an asse ssment of the support for each.

PAGE 164

! 150 Table 5.1 Summary of Support for Chapter Three Hypotheses Chapter 3 Hypotheses Support H 1 : Policy actors with more extreme policy views are more likely to reinforce their beliefs than change their beliefs Supported by the R esults More extreme views were associated with policy actors being more likely to report policy learning and that learning was more likely to be in the form of belief reinforcement and less likely to be in the form of a balance of belief ch ange and belief reinforcement. H 2 : Policy actors that seek advice more frequently from those with similar beliefs are more likely to reinforce their beliefs, rather than change their beliefs Not S upported by R esults Contradictory significant effects on policy learning as belief reinforcement or mostly reinforcement in only one of six policy belief quest ions. H 3 : Policy actors that collaborate more frequently with those with dissimilar beliefs are more likely to change their beliefs, rather than reinforce their beliefs Support Mixed More collaboration with those with dissimilar beliefs was associated with policy actors being less likely to report mostly belief reinforcement in two of six poli cy belief questions, but also was associated with policy actors being more likely to report no policy learning. H 4 : Policy actors that have participated in more frequent facilitated consensus based processes are more likely to change their beliefs, rather than reinforce their beliefs Not S upported by R esults Policy actor participation in facilitated consensus based processes had no effect on policy learning. The results indicate that : 1) e xtreme beliefs are associated with policy learning and also with increased belief reinforcement as the outcome of learning and 2) increased collaboration with individuals with differing policy views is associated with increased belief change CHAPTER FOUR RESEARCH QUESTION, HYPOTHESES, AND SUMMARY FINDINGS The objective of the analysis presented in Chapter Four was to examine fact ors shaping political learning. Research Question: What factors affect policy actor political learning in a local and state level energy and clima te policy subsystem?

PAGE 165

! 151 Table 5.2 below provides a summary assessment of the support for each of the Chapter Four hypotheses. Table 5.2 Summary of Support for Chapter Four Hypotheses Chapter 4 Hypotheses Support H1: Policy actors with extreme beliefs will be more likely to change or reinforce advocacy strategies than not change or reinforce advocacy strategies. Supported by R esults Policy actors with more extreme views were more likely to report political learning regarding advocacy strategies in general, and that learning was most likely to be in the form of reinforcement of advocacy strategies H2: Policy actors that participate in coalition building will be more likely to change or reinforce advocacy strategies than policy actors that do not par ticipate in coalition building. Not S upported by R esults Participation in coalition building had no affect on promoting political learning. H3: Policy actors that participate in facilitated multi stakeholder consensus based processes will be more likely to change or reinforce advocacy strategies than policy actors that do not participate in negotiated multi stakeholder consensus based processes. Not S upported by R esults Participation in facilitated multi stakeholder consensus based processes had no affect on promoting political learning. H4: Policy actors that participate in coalition building will be more likely to report advocacy strategy change than advocacy strategy reinforcement. Not S upported by R esults Participation in coalition building had no affect on shaping political learning. H5: Policy actors that participate in facilitated multi stakeholder consensus based processes will be more likely to report advocacy strategy change than advocacy strategy reinforcement. Not S upported by R esults Participati on in facilitated multi stakeholder consensus based processes had no affect on shaping political learning. The results indicate that extreme beliefs are associated with political learning, and with reinforcement of advocacy strategies as the outcome of learning SYNTHESIZED FINDINGS AND DISCUSSION The main goal of this dissertation was to examine factors that shape policy and political learning in policy actors, specifically to understand the role of existing beliefs and policy activities in shaping learning in the policymaking process. Belief c hange and reinforcement were included in the conceptualization of policy learning and change and reinforcement of behaviors were included in the definition of political learning.

PAGE 166

! 152 Before di scussing the major findings of the analysis across all chapters of this dissertation, discussion of some of the analysis this dissertation buil t upon is needed. Analysis by Elgin and Weible (2013) on the same sample of policy actors used in this study foun d that two advocacy coalitions indeed existed in the climate and energy policy subsystem of Colorado. Based on their findings, as well as the analysis presented in Chapter Two of this dissertation, it seems the coalitions did have the policy capacity and t he technical analytical resources hypothesized by the ACF to be necessary to engage in debate fostering learning in the subsystem (Elgin & Weible, 2013; Jenkins Smith et al., 2014). Further, a majority of policy actors in this coalition reported they engag ed in coalition building and facilitated consensus based processes with multiple stakeholders. This would seem to indicate the policy system could foster both within coalition learning and cross coalition learning (Sabatier, 1988; Jenkins Smith et al., 201 4) 18 Indeed, the findings presented in Chapter Two show policy actor learning is more common than nonlearning. The analysis in Chapter Three further explored policy learning and found that policy belief reinforcement was more common than belief change. Further exploration into political learning in Chapter Four contained findings showing that, in terms of political learning advocacy strategy reinforcement was also more common than advocacy strategy change. !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! !!!!!!!!!!!!!!! 18 It is worth noting, other factors hypothesized by the ACF to foster learning are: an intermediate level of informed con flict, conflict at the secondary aspects of belief systems (as opposed to core beliefs), and problems involving natural systems (as opposed to purely social systems ) While these factors were not examined directly in this study, climate and energy policy i s a natural system and the policy belief differences between the coalitions would seem to indicate some conflict, though the exact extent of conflict could not be stated with certainty absent more evidence.

PAGE 167

! 153 In terms of factors affecting policy and politi cal learning, t hree major findings emerge d from the analysis presented in C hapters Two and Three of this dissertation F irst extreme beliefs were shown to be associated with policy learning and with belief reinforcement (as opposed to belief change) as the product of policy learning. Second, extreme beliefs were also associated with political learning and with advocacy strategy reinforcement as the product of political learning. Finally o f the various policy activities measured, increased collaboration with policy actors with different views showed mixed results of increasing policy belief change, but most of the policy activities examined had little to no effect on policy or political learning. In summary, the results of this dissertation reflect some p redictions of the ACF and expand our understanding of the role of reinforcement in learning First, the policy actors sampled var ied across many attributes inclu ding organizational affiliation, frequency of engagement in different policy activities, and policy beliefs held. This aligns with the assumption of the ACF, built on the idea of a policy subsystem (Heclo, 1978) and a bottom up model of policy implementation (Hjern & Porter, 1981) that the group of actors active in subsystem would likely be diver se. Policy actors in the sample represented multiple levels of government agencies, private companies (e.g. in the energy sector ) non profit organizations such as environmental groups, and members of academic/research institutions. The degree to which th ese actors participated in various policy activities such as collaboration with other actors with dissimilar view s varied as did their reported sources of policy and political learning. Secondly, the ACF predicts policy actors with more extreme beliefs ma y be more constrained and less likely to learn. This fits into the broader description of boundedly

PAGE 168

! 154 rational actors in the policy process affected by preexisting values decision predicted by multiple frameworks used to understand the policy process (Cairne y & Heikkila, 2014). For example, both the ACF and the Narrative Policy Framework would predict extreme beliefs might cause policy actors to make decisions on policy issues based on the existing social construction of their place in a story of opposing coa litions, in turn leading to confirmation bias and reinforcement of beliefs and behaviors. This dissertation differentiated between policy learning and political learning and measured both types of learning as change of policy beliefs and political behavio rs, reinforcement of beliefs and behaviors, and the absence of either change or reinforcement. Given this conceptualization the findings indicate more extreme beliefs are associated with policy and political learning and that learning in both cases is mo re likely to be in the form of belief reinforcement and reinfo rcement of advocacy strategies. In terms of policy learning, extreme beliefs are associated with policy learning, and specifically associated with belief reinforcement as the outcome of learning In other words, policy actors with extreme beliefs are more likely to learn, and more likely to have their policy views reinforced, that is better supported or firmly rooted, as opposed to changed. However, this does not necessarily lead to the conclusi on that having extreme beliefs lead policy actors' to continue changing their beliefs by becoming more extreme. Because, as conceptualized by this study, belief reinforcement is not necessar ily the opposite of change (toward moderation or balance), but it is not unchanged (because nonlearning was indicated by neither change nor reinforcement ). More precisely then the findings here indicate having extreme beliefs, as compared to moderate belie fs, means policy actors will be more likely to learn (less likely to not learn). Secondly that, with

PAGE 169

! 155 extreme beliefs, the learning outcome is also more likely to be a more firmly held policy belief Finding s here then improve our understanding of reinforce ment in policy learning. E xtreme beliefs were shown in this study to have the effect of an anchor on policy actors' beliefs. Policy actors with extreme beliefs are then only more likely t o experience policy learning as further buttressing and supporting th eir policy beliefs by experience. R esults presented in Chapter Four allow further exploration of factors that shape individual political learning. Extreme policy beliefs are associated with a greater likelihood of political learning (in addition to policy learning), but also are associated with a specific product of political learning, behavioral reinforcement. In other words, policy actors with extreme beliefs are more likely to politically learn (either change or reinforce their advocacy strategies) The reason behind this may be because political learning is strategic in nature, and policy actors with more extreme beliefs have an increased desire to achieve policy goals, and thus a greater incentive to politically learn. The precise reasons why the produc t of learning was more likely to be advocacy strategy reinforcement though cannot be ascertained by the findings of this study. It may be that policy actors with more extreme views, in addition to experiencing more policy belief reinforcement are either a ctually more successful policy advocates (or just think they are) and thus only reinforce their advocacy strategies in a rational attempt to further policy goals. In either case, h opefully these findings can inform future efforts to further improve models and hypotheses of individual political lear ning related to reinforcement. Finally the ACF postulates cross coalition activities may be positively associated with lea r ning. But the reverse is also postulated. The insular political networks observed in these findings based on policy activities may be related to the so called devil shi ft,

PAGE 170

! 156 when policy actors will exaggerate the negative motives and influence of their opponents and avoid seeking advice or collabo rating with them but the devil shift was not explicitly examined in this study (Weible, Sabatier & McQueen 2009). The connection between the devil shift and policy intransigence has been shown in the ACF and the Narrative Policy Framework (Sabatier, Hu nter & McLaughlin, 1987). Again, given this study's conceptualization of learning the findings give some support to claim that collaborating with policy actors with different views and policy goals may mitigate reinforcement (firming rooting) of policy be liefs. Analysis here is not suggesting belief change (consensus, moderation, etc.) is the only pathway to policy change, nor is belief reinforcement necessarily a negative outcome. In any case, m ore evidence would need to be gathered to directly link this connection between cross coalition activities and policy learning as mitigated belief reinforcement, to either the concept of the devil shift or policy change. LIMITATIONS In terms of limitations of the study, of primary significance is that it is a s ingle study of one policy subsystem. Findings here may not be generalizable to other subsystems with different geographic, policy substantive scope, or policy actor populations. That said, this study's findings might be generalizability to other subnational climate and energy policy subsystems given Colorado's CAP comparability to other typical U.S. state Climate Action Plans, and Denver's comparability to typical U.S. Mayor's Climate Protection Agreement signatories T his subsystem could be considered a typi cal case among subnational climate and energy policy subsystems.

PAGE 171

! 157 This leads to the question as to how generalizable the findings of this study might be to policy subsystems other than a subnational climate and energy policy subsystem Besides the policy co ntent, some general descriptions of this subsystem might be useful to answer this question. This subsystem was shown to have two advocacy coalitions with differing policy beliefs. And t he policy actors in these coalition s w ere overwhelmingly homogenous ( wh ite not of Hispanic origin ) and well educated. Further, besides differing policy belief, these two coalitions were shown to be relatively similar in attributes of respective policy actor members. Both coalitions had similarly high levels of policy capacit y and engaged in similar policy activities at similar rates. Specifically, most policy actors in this subsystem engaged in coalition building and multi stakeholder consensus based forums. Other policy subsystems matching these generable descriptions may le nd themselves to more comparabil ity in the results found here. A related threat to external validity is any potential selection bias. T he s nowball sampling aspect of sample selection design migh t have fostered selection bias in terms of respondents suggest ing other policy actors with similar viewpoints, but this was mitigated to the extent that other target population identification and recruitment strategies were employed. The researcher feels confident that, to the extent that the target population of pol icy elites displays diversity (for instance in policy preferences, but less so in race/ethnicity), that diversity was captured by the sample. As always with self reported survey data there is a danger in threats to internal validity such as measurement err or. For instance perhaps some policy actors misunderstood questions, or perhaps they provided answers that were less than truthful. This first issue was addressed with the small beta test and resulting feedback conducted

PAGE 172

! 158 on the survey questions. But s elf reported measures of learning are potentially limited by other factors such as the memory of respondents and perceived social desirability of some answers. Relatedly, it is possible the measure of neutral parties as sources of learning is an imprecise metric for the concept for policy brokers in the ACF. Additional possibilities for future research survey questions are presented below. CONTRIBUTIONS Learning is a critical aspect of many approaches to the study of policy process. However, learning remain s an understudied concept due to the difficulty in parsimoniously examining different dimensions of learning processes and products ( Heikkila & Gerlak, 2013 ). The synthesized findings of this dissertation illuminate different factors affecting both policy and political learning. Some factors were shown to promote learning and to shape t he products of po licy and political learning as either belief and behavior change, or as belief and behavioral reinforcement. This study contributes to the literature on publ ic policy process theories in that, i n some cases it was building on existing ACF hypotheses related to policy learning. In other cases, new hypotheses relate d to political learning were employed here To start a discussion of this study's contributions, this is another application of theories of policy actor learning which broadens the scope and scale of applications, thereby adding to the literature testing existing theories and hypotheses. In terms o f this study's specific contributions to different ACF theoretical emphasis hypotheses, findings do not support some established ACF hypotheses related to advocacy coalitions. Specifically, this study found no evidence of a relationship between extreme bel iefs and organizational affiliation or a relationship between belief flexibility and organizational

PAGE 173

! 159 affiliation, which are currently included in established ACF hypotheses. A current ACF hypothesis related to coalition members and beliefs postulates that administrative agencies will usually advocate more moderate positions than their interest group allies." (Jenkins Smith et al., 2014, p. 195). While the unit of analysis for this study was individual policy actors, as opposed to agencies, there was no diff erence in the distribution of moderate beliefs (non extreme beliefs) between government employees and non government employees. Another established ACF hypotheses related to coalition members and beliefs states "Actors within purposive groups [e.g. envir onmental nonprofit organizations] are more constrained in their expression of beliefs and policy positions than actors from material groups [e.g. private sector energy business]" (Jenkins Smith et al., 2014, p. 195). This study found policy actors' extrem eness of belief and belief flexibility does not vary by organizational affiliation. It is possible this hypothesis does not hold true in every policy subsystem, but it is also possible that by grouping all policy actors from nonprofit organization s (some o f which may not be ideological purposive groups) together and all private sector employees (which may not all be from energy companies) together, the difference in expression of beliefs was obscured here. A lesson drawn from this study is that future ACF a pplications attempting to test these two ACF hypotheses may want to consider further refining the organization affiliation survey question to increase granularity of responses related to purposive and material groups. Jenkins Smith et al. ( 2013) call for i ncreased attention in ACF applications to examine the factors that shape individual level learning as belief change or resistance to change In terms of this study's contributions to established ACF hypotheses related to

PAGE 174

! 160 learning, findings here would suggest extreme policy beliefs and, to a lesser extent, cross coalition collaboration, are important factors in individual level conative policy l earning and behavioral political learning processes. But the findings of this study contribute by examining both learning and nonlearning, and allow for further examination of an important existing ACF argument. Specifically, ACF theory claims, "Given the importance of belief systems in filtering and interpreting information, for example, the expectation is that coalition actors with extreme beliefs will be less likely to learn from opponents than coalition actors with more moderate beliefs" (Jenkins Smith et. al., 2014). Results of this study show extreme beliefs are associated with learning, and with learning as belief reinforcement, but advice seeking and collaboration with opponents had association alone with learning or belief reinforcement, or any no i nteraction effect with extreme belief. Therefore policy learning is not constrained by extreme beliefs ; in fact policy learning is promoted by extreme beliefs, but that learning is most likely to the form of belief reinforcement. This finding comports wit h previous ACF findings that beliefs are resistant to change (Jenkins Smith et al., 2014), and potentially with the broader social science observation that individuals selectively filter information (Lord et al., 1979; Lindblom 1959; Simon, 1945). However results of this study indicate a similar association between extreme beliefs and political learning, and an association specifically between extreme beliefs and advocacy strategy reinforcement, but cross coalition activities did not have the same mitigat ion of reinforcement effect in policy actors' political learning as it did with policy learning. In other words, some cross coalition activities (but interestingly not participation in forums such as facilitated consensus based processes) mitigated policy

PAGE 175

! 161 learning as belief reinforcement, but not political learning as advocacy strategy reinforcement. This is perhaps one of this study's greatest contributions to ACF theory. Namely, cross coalition activities were shown in this study to affect policy learning differently than they did political learning. Early ACF theory and augments define learning as including belief and behaviors (Sabatier, 1988; Jenkins Smith & Sabatier, 1993), but established ACF learning related hypotheses do not differentiate between po licy and political learning (Weible, Sabatier & McQueen, 2009; Weible et al., 2011). This study's finings indicate it is potentially worth exploring separate hypotheses for policy and political learning. Another contribution this ACF application offers is that within coalition learning related activities (e.g. coalition building, seeking advice from those with similar beliefs) and that cross coalition activity such as facilitated consensus based processes had no effect on either policy or political learnin g. That is, these activities were not associated with either policy or political learning, or associated with any specific learning outcome (e.g. belief reinforcement or advocacy strategy change). This would indicate that hypothesized relationships betwee n learning and these policy activities are not present in every policy subsystem. Perhaps the energy and climate policy subsystem is sufficiently polarized in terms of different policy beliefs, thus these cross coalition activities could not affect learnin g. It may be fruitful to explore, through additional ACF applications, if the lack of relationship between cross coalition policy activities and learning is also found in other subnational climate and energy policy subsystems. It is possible no associations between participation in professional forums and policy or political learning were found in the analysis of this study because the survey

PAGE 176

! 162 questions could be more specific. In other words, perhaps questions related to forums (e.g. facilitated consensus based processes), which explicitly name the specific forum (for example: the Colorado Climate Action Plan Roundtable Workshops) would tie learning processes to policy actors' experiences and activities more directly and intuitively. By linking me mories of specific experiences, in specific forums, to changes or reinforcement of beliefs and advocacy tactics, as opposed to generally asking about "roundtables perhaps policy actors might report learning differently. Further, this study 's examination of change and reinforcement of belief and behavior as distinct forms of learning goes a step further than simply examining change and resistance to change. Meta analysis of ACF applications has shown that learning related hypotheses are among the most explicitly tested of the formal hypotheses (Weible, Sabatier & McQueen, 2009). Though current established ACF learning related hypotheses do not specifically differentiate belief or behavioral reinforcements as learning products (Jenkins Smith et. al., 201 4). That is, May's (1992) policy learning and political learning are not treated differently in existing ACF hypotheses, though they are both implicitly included in the original ACF theoretical definition "policy oriented learning This study was one of t he first attempts to determine if policy actor attributes (e.g. extreme beliefs and policy activities) affected policy and political learning differently. Hopefully future applications can further explore how political and policy learning are shaped diffe rently based on different factors and mechanisms in other policy subsystems. T he findings here may help refine ACF and learning related hypotheses but this study is also part of a larger research agenda aimed at improving

PAGE 177

! 163 models of individ ual learning in the ACF as well as other policy process theories and frameworks in general. For instance, scholars working in the field of studying Social Ecological Systems are ultimately trying to find pattern s than can improve policy related to sustainable resource use T o do so we must learn and we must understand factors that affect learning ( Schewenius, McPhearson & Elmqvist 2014 ; Anderies, 2015 ). Anderies & Janssen (2013) specifically call on local level experimentation and learning as essential elements of sustainable public policy and for research and theorizing focused on outcomes of the policy process other than policy change. While this dissertation hopes to be one additional small step toward this rather grandiose cause it should be noted that learni ng is not assumed here to lead to improved or even new policy outcomes. That said, as scholars examining learning in the context of adaptive governance have pointed out, understanding factors that may p romote learning is essential to the efforts in collabo rative and sustainable policymaking (Lubell, 2004; Henry, 2009). This study contributes to these efforts in providing insight as to which factors promote individual learning in this subsystem, and lead to different forms of learning. No evidence was found in the findings of this study that policy activities such as facilitated consensus based processes promote learning, but future exploration could further attempt to link extreme beliefs and policy actors' social networks to learning outcomes. This study co ntributes to thinking through future potential hypotheses related to different learning outcomes such as change and reinforcement of beliefs and advocacy strategies based on different social networks. Building on the results here, which found differences i n ways some factors effect policy and political learning, scholars could attempt to parse

PAGE 178

! 164 the differences between policy and political learning in theorizing regarding social learning (Reed et al., 2006) and collective learning processes (Heikkila & Gerlak 2013). In the context of the projected consequences of climate change, understanding learning processes at the subnational governance level may be especially important. According to Bulkeley (2013) the patchwork of climate policy innovation at the subna tional level provides critical lessons for adaptive governance. In many ways these subsystems are climate policy experiment s which, in the absence of more substantial national policy, are attempts to problem solve with new policy designs, and thus through incremental trial and error, collective learning as part of adaptive governance is possible (Hoffman, 2011). This study cont ributes to efforts to understand these learning processes by highlighted factors important in shaping learning at the individual level of climate and energy policy actors. Better understanding factors that foster or prevent learning from these policy exper iments, successful or otherwise, is crucial to our ability to respond to climate change at the local, national and global governance level (Bulkeley, 2013). Recently Stern and Dietz (2015) specifically called out the Intergovernmental Panel on Climate Cha nge to foster more engagement of social scientists, along with the hard working ecologists and physicists modeling climate change and associated problems, in their processes Doing so would give the opportunity for social concepts and methods such as thos e explored here to be applied toward benefitting society regarding improving the human environmental interactions involved in climate change policy and energy issues. Findings here suggest which factors might be responsible for shaping climate and energy policy actors' learning in different ways, which can be tested by other researchers using other samples of similar policy actors in different policy subsystems.

PAGE 179

! 165 This study att empted to clearly conceptualize and measure policy and political learning in nove l ways that were built on the existing literature. For instance, this study was one of the first that s pecifically included reinforcement and change (in beliefs and behaviors) as products of learning in its conceptualization of policy and political learnin g. This is important because while the concept of reinforcement of beliefs and behaviors is included in some conceptualizations of learning in the policy process, the two dimensions of learning rarely, if ever, are measured together. This study provides s ome insight into the ways different factors might shape learning in very different directions, which might have consequences for possible policy change or stalemate over time. For instance, more belief and advocacy strategy reinforcement may, in some cases lead to a lack of policy change due to a dialogue of the deaf. This is not to suggest belief or advocacy strategy change is a more desirable outcome of learning, or that belief and strategy change is the only pathway to policy change. In addition to measuring both reinforcement and change of policy beliefs and advocacy strategies, this study measured the absence of either (non learning) another relatively novel contribution Hopefully this study is useful to future studies working to re fine the way learning is examined in the policy process literature. Specifically, understanding contextual factors that encourage or discourage learning in the policy process would seem important given the need for adaptive leaning in human ecological syst ems such as the climate and our relationship to it. Relatedly, this study was one of the first to measure political learning by directly asking about changes (and reinforcements) to advocacy behavior. Ideally, t hese finding may help advance the study of po litical learning in addition to policy learning in the ACF, as well as in other policy process

PAGE 180

! 166 theories in that the direct and quantitative methods used here may shed additional light on concepts typically exp lored in less systematic ways. FUTURE RESEARCH One direction future research could be conducted would be to endeavor to break up the policy actors into various sub groups (e.g. local government versus state government actors, or individuals from traditional environmental non profit organizations versus energy specific related non profit organizations) to potentially tease out differences in policy actor policy or political learning based on demographic factors. Some of this could potentially be done with this data set with some additional research but a comparative sample asking similar questions re g a r ding learning would likely be preferable because it would also provide cross case comparison as well as more granula rity in demographic attributes. Another potential extension of this study for futur e research agenda is that it is important to understand how aggregations of individuals in coalitions or organizations may learn at the collective level. This is a theoretical argument that could be taken up by f uture analysis of policy actors from this, o r another subsystem, asking similar learning questions Individuals could be aggregate d into groups (e.g. by organizational affiliation or advocacy coalition) to examine if similar relationships between independent variables and policy and political learn ing exist at the collective learning level. In words, groups could be included as an additional variable in the exam ination of drivers of learning. If this study was to be repeated or improved upon, perhaps some refinements could be made to the survey ques tions (the measures) used. Certainly there are many ways learning could be measured. Close d ended survey questions using phrases such as

PAGE 181

! 167 how much have you learned directly are at least one option. An ordinal or interval scale of learning would have allowed for additional statistical tests to be employed in order to test the hypotheses. This is complicated however by the concept of nonlearning, and the extent to which this concept corresponds to a zero on an interval scale for instance Relatedly, a lack of belief/behavioral change and/or reinforcement may not be the only way to measure the concept of nonlearning Perhaps policy actors could be asked to what extent have you learned with not at all as a possible response. In terms of t he two types of learning examined in this dissertation, at least two other lessons learned may help inform future survey question design. First, the sources of policy learning (belief change or reinforcement) could have been examined more directly in a mor e useful way. The extent to which neutral parties, members perceived to not be a member of any advocacy coalition, influence the actors' policy learning (as either belief change or reinforcement) would be desirable. Specifically, the theoretical construct of policy brokers in the ACF literature could be reexamined for words and phrases to be used directly in future survey questions to measure this variable. In other words, perhaps there are direct ways to measure policy brokers that could be explored in future research. Similarly, survey questions could have made a more direct link between policy actors' perceived success (or failure) of advocacy strategies to their political learning. In other words, future studies could specifical ly examine whether inde ed perceived successful advocacy strategies are more likely to be reinforced, while perceived ineffective advocacy strategies are more likely to be changed. This would help explore

PAGE 182

! 168 the causal mechanism behind different outcomes of political learning which could not be examined her e. In closing, c limate change has emerged as the defining environmental challenge of our time. This dissertation advances our understanding of how climate and energy policy actors learn and can contribute constructively to commun ities of policy practitioners in this field. By adding to theory building and measuring efforts of individual learning products we can better understand the role of learning in different policy processes as well as the factors that shape learning. Likewi se, by better understanding different climate and energy policy processes we can better understand which of these contexts may be more beneficial to society. This researcher is optimistic that policy process research can improve policy outcomes. It is af ter all not nobler to suffer the slings and arrows of outrageous academia than it is to take up arms against a sea of greenhouses gasses, and by opposing mitigate them. We must do both.

PAGE 183

! 169 R EFERENCES Albright, E. A. ( 2011 ) Policy Change and Learning in Response to Extreme Flood Events in Hungary: An Advocacy Coalition Approach. Policy Studies Journal 39 (3), pp. 485 511. Albright, E. A. & Crow. D. A. ( 2014 ). Democratic Process, Resource Shifts, and Policy Learning: Analyzing Responses to Colorado's Extreme Flood Events of 2013. Paper Presented at the Association for Public Policy and Management November 2014 Conference. Anderies, J. M. (2015). Understanding the Dynamics of Sustainable Social Ecological Systems: Human Behavior, Institutions, and Reg ulatory Feedback Networks. Bulletin of Mathematical Biology (77), pp. 259 280. Anderies, J. M. & Janssen, M. A. (2013). Robustness of Social Ecological Systems: Implications for Public Policy. Policy Studies Journal, 41 (3), pp. 513 536. Bailey, J. ( 2007 ) Lessons from the Pioneers: Tackling Global Warming at the Local Level Washington D.C.: Institute for Local Self Reliance. Balla, S. ( 2001 ). Interstate Professional Associations and the Diffusions of Policy Innovations. American Politics Research 29 (3), pp. 221 245. Bandelow, N. ( 2006 ). Advocacy Coalitions, Policy Ori ented Learning, and Long Term Change in Genetic Engineering Pol icy: An Interpretist View. German Policy Studies 3 (4), pp. 747 805. Bennett, C. J. & Howlett, M. ( 1992 ) The Lessons of Learning: Reconciling Theories of Policy Learning and Policy Change. Policy Sciences 25 pp. 275 94. Berry, F. S. & Berry W. D. ( 2014 ) "Innovation and Diffusion Models in Policy Research." I n P. A. Sabatier and C. M. Weible (Ed s .) Theories of the Policy Process (3 rd ed.) Boulder, CO: Westview Press. Betsill, M. ( 2001 ). Mitigating Climate Change in US Cities: Opportunities and Obstacles. Local Environment 6 (4), pp. 393 406. Birkland, T. A. ( 2011 ) An Introduction to the Policy Process (3 rd ed.). Armonk, NY: M.E. Sharpe, Inc. Bulkeley, H. ( 2013 ) Cities and Climate Change New York, NY: Routledge. Bulkeley, H. & Betsill. M. ( 2003 ) Cities for Climate Change: Urban Sustainability and Global Environmental Governance London, UK: Routledge.

PAGE 184

! 170 Busenberg, G. J. ( 2001 ) Learning in organizations and public policy. Journal of Public Policy 21 (2), pp. 173 189. Byrne, J., Hughes, K., Rickerson, W., & Kurdgelashvili, L. ( 2007 ). American policy conflict in the greenhouse: Divergent trends in federal, regional, state, and local green e nergy and climate change policy Energy Policy 35 pp. 4555 4573 Butler, D. M., Volden, C., Dynes, A. M., & Shor, B. (2015). Ideology, Learning, and Policy Diffusion: Experimental Evidence. American J ournal of Political Science, XX (Xx), pp. 1 13. Cairney, P. & Heikkila, T. ( 2014 ) "A Comparison of Theories of the Policy Process." I n P. A. Sabatier and C. M. Weible (Ed s .) Theories of the Policy Process (3 rd ed.) Boulder, CO: Westview Press. The City and County of Denver. ( 2006 ) The Greenprint Den ver Plan. Churkina, G. ( 2008 ). Modeling th e Carbon Cycle of Urban Systems Ecological Modeling 216 pp. 107 113. Clark, W. C. ( 2007 ). Sustainabi lity science: A room of its own Proceedings of the National Academy of Sciences 104 (6), pp. 1737 1738. Crona B. I. & Parker, J. N. ( 2012 ). Learning in support of governance: Theories, methods, and a framework to assess how bridging organizations contribute t o adaptive resource governance. Ecology and Society 17 (1), p. 32. Crossan, M. N., Lane, H. W., & White R. E ( 1999 ). An Organizational Learning Framework: From Intuition to Institution. Academy of Management Review 24 (3), pp. 522 37. Davis, M. D. & Weible. C. M. ( 2011 ). Linking Soci al Actors by Linking Social Theories: Towards Improved GHG Mitigation Strategies Carbon Management 2 (4), pp. 483 491. d e Leon, P. ( 2006 ) The historical roots of the field In M. Moran, M. Rein, & R. E. Goodin (Eds.), The Oxford Handbook of Public Policy Oxford, UK: Oxford University Press. d e Leon, P. ( 1997 ) Democracy and the Policy Sciences Albany, NY: State University of New York Press. Dietz, T., Ostrom, E., & Stern P. ( 2003 ). The Struggle to Govern the Commons. Science 302 pp. 1907 1912. Easton, D. (1953). The Political System: An Inquiry Into the State of Political Science New York, NY: Alfred A. Knopf.

PAGE 185

! 171 Eberg, J. ( 1997 ) Waste Policy and Learning: Policy Dynamics of Waste Management and Waste Incineration in the Netherlands and Bavaria. Delft, the Netherlands: Uitgeverij Eburon. Elgin, D J. & Weible, C. M. ( 2013 ). A Stakeholder Analysis of Colorado Climate and Energy Issues Using Policy Analytical Capacity and th e Advocacy Coalition Framework. Review of Policy Research 30 (1), pp. 114 132. Elgin, D., Pattison, A., & Weible, C. M. ( 2012 ) Policy Analytical Capacity Inside and Outside Government: A Case Study of Colorad o Climate and Ene rgy Issues. Canadian Political Science Review 6 (1), pp. 101 116. Elliot, B. A. & Schlaepfer, R. (2001). The Advocacy Coalition Framework: Application to the Policy Process for the Development of Forest Certification in Sweden. Journal of European Public Policy 8(4), pp. 642 641. Ellison, B. A. ( 1998 ). The Advocacy Coalition Framework and Implementation of the Endangered Species Act: A Case S tudy in Western Water Politics. Policy Studies Journal 26 (1), pp. 11 29. Environmental Protection Agency. ( 2015 ) State Climate and Energy Program: State Examples Climate Change Action Plans. http://www.epa.gov/statelocalclimate/state/ s tate examples/action plans.html Last accessed: 4/29 /15 Freeman, R. ( 2006 ) "Learning in Public Policy In Moran, M., Rein, M., & Goodin R. E. (Eds.), The Oxford Handbook of Public Policy Oxford, UK: Oxford University Press. Fogel, C. ( 2007 ) "Constructing Progressive Climate Chan ge Norms: The US in the Early 2000s." In Pettenger, M. E. (Ed.). The Social Construction of Climate Change Burlington, VT: Ashgate Publishing Limited. Fowler Jr., F. ( 2002 ) Survey Research Methods, 3 rd Ed. Thousand Oa ks, CA: Sage Publications, Inc. Fry, B. R. & Raadschelders, J. C. N. ( 20 08 ) Mastering public administration: From Max Weber to Dwight Waldo, 2 nd Ed. Washington D.C.: CQ Press. Geor ge, A. L. & Bennett, A. ( 2005 ) Case Studies and Theory Development in the Social Sciences Cambridge, MA: MIT Press. Gerlak, A. K. & Heikkila, T. ( 2011 ). Building a Theory of Learning in Collaboratives: Evidence from the Everglades Restoration Program. Journal of Public Administration Research and Theory 21 pp. 619 644. Gerring, J. ( 2007 ) Case Study Research: Principles and Practices Boston, MA: Cambridge University Press.

PAGE 186

! 172 Giddons, A. ( 2011 ) The Politics of Climate Change, Second Edition Cambridge, MA: Polity Press. Glaeser, E. L. & Kahn, M. E. (2008) The Greenness of Cities: Carbon Dioxide Emissions and Urban Development. Journal of Urban Economics 67 pp. 404 418. Goertz, G. ( 2006 ) Social Science Concepts: A User's Guide Princeton, NJ: Princeton University Press Griebler, E. & Hadolt, B. ( 2006 ). Policy Learni ng in Policy Domains with Value Conflicts: The Austrian Case of Abortion and Assi sted Reproductive Technologies. German Policy Studies 3 (4), pp. 698 746. Grin, J. & Loeber, A. ( 2007 ) "Theories of Policy Learning: Agency, Structure, and Change I n Fischer, F., Miller, G. J., & Sidney, M. S. (E ds. ), Handbook of Public Policy Analysis Boca Raton, FL: CRC Press. Heclo, H. ( 1974 ) Modern Social Politics in Britain and Sweden: From Relief to Income Maintenance. New Haven, CT: Yale Universit y Press. Heclo, H. (1978). "Issue Networks and the Executive Establishment In A. King (Ed.) The New American Political System Washington, D.C.: American Enterprise Institute. Heikkila, T. & Gerlak, A. K. ( 2013 ). Building a Conceptual Approach to Collective Learning: Less ons for Public Policy Scholars. Policy Studies Journal 40 (3), pp. 484 512. Heikkila, T., Pierce, J. J., Gallaher, S., Kagen, J., Crow, D. A., & Weible, C. M. ( 2014 ). Understanding a Period of Policy Change: The Case of Hydraulic Fracturing Discl osure Policy in Colorado. Review of Policy Research 31 (2), pp. 65 87. Henry, D. A. ( 2009 ). The Challenge of Learning for Sustainabi lity: A Prolegomenon to Theory. Research in Human Ecology 16 (2), pp. 131 140. 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), pp. 419 444. Hillman, T. & Ramaswami, A. ( 2010 ). Greenhouse gas emission footprints and energy use benchmarks for eight US cities. Environmental Science & Technology 44 pp. 1902 1910. Hirsch, R., Baxter, J. & Brown, C. (2010). The Importance of Skillful Community Leaders: Understanding Municipal Pesticide Policy Change in Calgary and Halifax. Journal of Environmental Planning and Management 53 (6), pp. 743 757.

PAGE 187

! 173 Hjern, B. & Porter, D. (1981). Implementation Structures: A New Unit of Administrative Analysis. Organization Studies, 2 pp. 211 227. H offmann, M. J. (2011). Climate Governance at the Crossroads: Experimenting with a Global Response after Kyoto Oxford, UK: Oxford University Press. Howlett, M. ( 2009 ). Policy analytic capacity and evidence based policy making: lessons from Canada. Canadian Public Administration 52 (2), pp. 153 1 75. Ingold, K. & Varone. F. ( 2012 ). Treating Policy Brokers Seriously: Evidence from the Climate Policy Journal of Public Adm inistration Research and Theory, 22 (2), pp. 319 346. Intergovernment al Panel on Climate Change. ( 2013 ). Climate Change 2013: The Physical Science Basis, Working Group I Contribution to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change Stockholm, Sweden : IPCC. Intergovernment al Panel on Climate Change. ( 2014 ). Climate Change 2014: Synthesis Report Stockholm, Sweden: IPCC. Jenkins Smith, H. C., Norhrstedt, D. Weible, C. M. & Sabatier, P. ( 2014 ) "The Advocacy Coalition Framework: Foundations, Evolution, and Ongoing Research." I n P. A. Sabatier and C. M. Weible (Ed s .) Theories of the Policy Process (3 rd ed.) Boulder, CO: Westview Press. Jenkins Smith, H. C. & Sabatier, P. A. (1993). "The Dynamics of Policy Oriented Learning." In. P. A. Sabatier and H. C. Jenkins Smith (Eds.), Polic y Change and Learning: An Advocacy Coalition Approach Boulder, CO: Westview Press. Jenkins Smith, H. C. Silva, C. L., Gupta, K. & Ripberger, T. (2014). Belief Continuity and Change in Policy Advocacy Coalitions: Using Cultural Theory to Specify Belief Systems, Coalitions, and Sources of Change. Policy Studies Journal, 42 (4), pp. 484 508. Jeon, Y. & Haider Markel, D. P. (2001). Tracing Issue Definition and Policy Change: An Analysis of Disability Issue Images and Policy Response. Policy Studies Journal, 29 (2), pp. 215 231. Kingdon, J. W. ( 2003 ) Agendas, Alternatives, and Public Policy, 2nd Edition New York, NY: Longman. Kraft, M. E. & Furlong, S. R. ( 2015 ) Public Policy: Politics, Analysis, and Alternatives, 5 th Edition Thousand Oaks, CA: Sage Publications, Inc. Krause, R. ( 2011 ). Policy innovation, intergovernmental relations, and the adoption of climate prote ction initiatives by U.S cities Journal of Urban Affairs 33 (1), pp. 45 60.

PAGE 188

! 174 Kubler, D. ( 2001 ). Understanding Policy Change with the Adv ocacy Coalition Framework: an Ap plication to Swiss Drug Policy. Journal of European Public Policy 8 (4), pp. 623 641. Lachapelle, E Borck, C. P. & Rabe, B. ( 2012 ). Public attitudes toward climate science and climate policy in federal systems: Canada and the United States compared. Review of Policy Research 29 (3), pp. 334 357. Larsen, J. B., Vrangbaek, K. & Traulsen, J. M. ( 2006 ). Advocacy Coalitions and Pharmacy Policy in Denmark. Social Science and Medicine 63 (1), pp. 212 24. Layzer, J. A. ( 2006 ) The Environmental Case: Translating Values Into Policy, (2 nd ed.). Washington, D.C. : CQ Press. Leach, W. & Sabatier, P. ( 2005 ). To Trust an Advers ary: Integrating Rational and Psychological Models of Collaborative Policymaking. American Political Science Review 99 (4), pp. 491 503. Leach, W. D., Weible, C. M., Vince, S. R., Siddiki S. N. & Calanni, J. (2013). Fostering Learning through Collaboration: Knowledge Acquisition and Belief Change in Marine Aquaculture. Journal of Public Administration Research and Theory 24 (3), pp. 591 622. Lester, J. & Hamilton, M. ( 1987 ) "Intergovernmental Relations and Marine Policy Change." In M. E. Silva (Ed.) Ocean Resources and U.S. Intergovernmental Relations in the 1980s Boulder, CO: Westview Press. Lindblom, C. E. ( 1959 ). The Science of Muddling Through Public Administration Review 19 (2), pp. 79 88. Litfin, K. ( 2000 ). Advocacy Coalitions Along the Domestic Foreign Frontier: Globalization and Canadian Climate Change Policy. Policy Studies Journal 28 (1), pp. 236 252. Lord, C. G., Ross, L. & Lepper, M. R. ( 1979 ). Biased assimilation and attitude polarization: The effects of prior theories on su bsequently considered evidence. Journal of Personality and Social Psychology 37 pp. 2098 2109. Long, J. S. ( 1997 ) Regression Models for Categorical and Limited Dependent Variables Thousand Oaks, CA: Sage Publications, Inc. Lon g, J. S. & Freese, J. ( 2014 ) Regression Models for Categorical Dependent Variables Using Stata, (3 rd ed) College Station, TX: Stata Press Publication. Lubell, M. ( 2003 ). Collaborative Institutions, Belief Systems, and Perceived Policy Effectiveness. Political Research Quarterly 56 (3), pp. 309 23.

PAGE 189

! 175 Lutsey, N. & Sperling, D. ( 2008 ). America's bottom up c limate change mitigation policy Energy Policy 36 pp. 673 685 Lyng, J. ( 2010 ). Colorado's 30% Renewable Energy Standard Policy Design and New Markets Denver, CO: Colora do Governor's Energy Office. MacQueen, J. ( 1967 ) "Some methods for classification and analysis of multivariate observations." In L. M. Le & J. Neyman (Eds.), Proceedings of the fifth Berkeley symposium on mathematical statistics and probability Berkeley CA : University of California Press. May, P. J. ( 1992 ). Policy Learning and Failure. Journal of Public Policy 12 (4), pp. 331 354 McBeth, M. K, Jones, M. D. & Shanahan, E. A. (2014). "The Narrative Policy Framework". I n P. A. Sabatier and C. M. Weible (Ed s .) Theories of the Policy Process (3 rd ed.) Boulder, CO: Westview Press. Meijerink, S. V. ( 2005 ) Understanding Policy Stability and Change: the Interplay of Advocacy Coalitions and Epistemic Communities, Windows of Opportunity, and Dutch Coa stal Flooding Policy 1956 2003. Journal of European Public Policy 12 (6), pp. 1060 10 77 Milner, B., Squire, L. R. & Kandel, E. R. (1998). Cognitive Neuroscience and the Study of Memory. Neuron, 20 (3), pp. 445 468. Mintrom, M. & Vergari, S. ( 1996 ) Advocacy Coalitions, Policy En trepreneurs, and Policy Change. Policy Studies Journal 24 (3), pp. 420 434. Mintrom, M. & Norman, P. ( 2009 ). Policy Entr epreneurship and Policy Change. The Policy Studies Journal 37 (4) pp. 649 667. Montpetit, E. ( 2012 ). Does Holding Beliefs with Conviction Prevent Policy Actors from Ad opting a Compromising Attitude? Political Studies 60 pp. 621 642. Munro, G. D. & Ditto, P. H. ( 1997 ) Biased assimilation, attitude polarization, and affect in reactions to stereotype relevant scientific information. Personality and Social Psychology Bulletin 23 (6), pp. 636 653. Munro, G. D., Ditto, P. H., Lockhart, L. K., Fagerlin, A., Gready, M., & P eterson, E. ( 2002 ) Biased assimilation of sociopolitical arguments: Evaluating the 1996 U.S. Presidential debate. Basic and Applied Social Psychology 24 (1) pp. 15 26. Munro, John. ( 1993 ) "California Water Politics: Explaining Change in a Cognitively P olarized Subsystem." In P. A. Sabatier & H. Jenkins Smith (Eds.) Policy Change and Learning Boulder, CO: Westview Press.

PAGE 190

! 176 Muro, M. & Jeff rey, P. ( 2008 ) A critical review of the theory and application of social learning in participatory natura l resource management processes Journal of Environmental Planning and Management 51 (3), pp. 325 344. Newell, J. P. & Cousins, J. J. (2014). The Boundaries of Urban Metabolism: Towards a Political Industrial Ecology. Progress in Human Geography Online December 22 pp. 1 26 Niederberger, A. ( 2005 ) Science for Climate Change Poli cy Making: Applying Theory to Pra ctice to Enhance Effectiveness. Science and Public Policy 32 (1), pp. 2 16. Nohrstedt, D. ( 2005 ). External Shocks and Policy Change: Three Mile Island and Swedish Nuclear Energy Policy. Journal of European Public Policy 12 (6), pp. 1041 1059. Nohrstedt, D. ( 2008 ). The Politics of Crisis Policymaking: Chernobyl and Swedish Nuclear Energy Policy." P olicy Studies Journal. 36 (2), pp. 257 278. Nohrstedt, D. ( 2011 ). Shifting Resources and Venues Producing Policy Change in Contested Subsystems: A Case Study of Swedi sh Signals Intelligence Policy. Policy Studies Journal 39 (3), pp. 461 84. Olsen, R. S., Olson, R. A. & Grawronski, V. T. (1999). Some Buildings Just Can't Dance: Politics, Life Safety, and Disasters S tamford, CT: Jai Press, Inc. Ostrom, E. ( 2005 ) Understanding Institutional Diversity Princeton, NJ: Princeton University Press. Ostrom, E., Cox, M., & Schlager, E. (2014). "An Assessment of the Institutional Analysis and Development Framework and Introduction of the Social Ecological Systems". I n P. A. Sabatier and C. M. Weible (Ed s .) Theories of the Policy Process (3 rd ed.) Boulder, CO: Westview Press. Rabe, B. ( 2004 ) Statehouse and Greenhouse: The Emerging Politics of American Climate Change Policy Washington, D.C. : Brookings Institutional Press. Ramaswami, A., Hillman, T., Janson, B., Reiner, M., & Thomas, G. ( 2008 ). A Demand Centered, Hyb rid Life Cycle Methodology for City Scale Greenhouse Gas Inventories. Environmental Science & Technology, 42 (17), pp. 6455 6461. Ramaswami, A., Bernard, M., Chavez, A., Hillman, T., Whitiker, M., Thomas, G. & Marshall, M. ( 2012a ). Quantifying Carbon Mitigation Wedges in U.S. Cities: Near Term Strategie s Analysis and Critical Review. Environmental Science & Technology 46 (7) pp. 3629 3642.

PAGE 191

! 177 Ramaswami, A. Weible, C., Main, D., Heikkila, T., Siddik, S., Duvall, A., Pattison, A., & Be rnard, M. ( 2012b ). A Social Ecological Infrastructural Systems Framework for Interdisciplinary Study of Sustainable City Systems. Journal of Industrial Ecology 16 (6), pp. 801 813. Ramseur, J. ( 2007 ). Climate Change: Action by States to Address Greenhouse Gas Emissions Washington, D.C.: Congressional Research Servic e. Reed, M. S., Evely A. C., Cundill, G., Fazey, I., Glass, J., Laing, A., Newig, J., Parrish, B., Prell, C., Raymond, C. & Stringer, L. C. ( 2010 ). What is social learning? Ecology and Socie ty 15 (4), pp. 1 10. Regan P. M. (2015). The Politics of Global Climate Change Bo ulder, CO: Paradigm Publishers. Ritter, B. ( 2007 ). Colorado Climate Action Plan: A Str ategy to Address Global Warming Denver, CO: Colorado Governor's Office. Rose, R. (1991). What is Lesson Drawing? Journal of Public Policy, 11 (1), pp. 3 30 Rose, R. (1993). Lesson Drawing in Public Policy. A Guide to Learning Across Time and Space Chatham, NY, Chatham House Publishing. Rubin, H. & Rubin, I. ( 2005 ) Qualitative Interviewing: the Art of Hearing Data Thousand Oaks, CA: Sage Publications, Inc. Sabatier, P. ( 1987 ). Knowledge, Policy Oriented Learning, and Policy Change An Advocacy Coalition Framework Knowledge: Creation, Diffusion, Utilization 8 (4), pp. 649 692 Sabatier, P. ( 1988 ). An advocacy coalition framework of policy change and the role of policy oriented learning therein. Policy Sciences 21 pp. 129 168. Sabatier, P. ( 1998 ). The advocacy coalition framework : revisions and relevance for Europe. Journal of European Public Policy 5 (1), pp. 98 130. Sabatier, P. ( 2007 ) "The Need for Better Theories" In Sabatier, P. A. (Ed.) Theories of the Policy Process, (2 nd ed.) Cambridge, MA: Westview Press. Sabatier, P. A. & Brasher, A. M. ( 1993 ). From Vague Consensus to Clearly Differentiated Coalitions: Environmental Policy at Lake Tahoe, 1964 1985." In Sabatier, P. & Jenkins Smith, H. (Eds.) Policy Change and Learning. Boulder, CO: Westview Press. Sabatier, P. A., Hunter, S. & McLaughlin, S. (1987). The Devil Shift: Perceptions and Misperceptions of Opponents. Western Political Quarterly, 41 pp. 449 476.

PAGE 192

! 178 Sabatier, P. & Jenkins Smith, H. ( 1999 ) "The Adv ocacy Coalition Framework: An Assessment." I n Sabatier, P. (Ed.) Theories of the Policy Process Boulder, CO: Westview Press. Sabatier, P. A. & Jenkins Smith, H. C. ( 1993 ) Policy change and learning: An advocacy coalition approach Boulder, CO: Westview Press. Sabatier, P. & W e i ble, C. ( 2007 ) "The Advocacy Coalition Framework: Innovations and Cla rifications." In Sabatier, P. A. (Ed.) Theories of the Policy Process, (2 nd ed.), Cambridge, MA: Westview Press. Sabatier, P., Weible, C., Hulsman, M. & Nechod om, M. ( 2003 ). Stakeholder Belief Change in th e Lake Tahoe Basin: 1970 2001 A Final Report to the USDA Forest Service, Pacific Southwest Research Sta tion. Berkeley, CA. Schneider, A. & Ingram, H. ( 1993 ). The Social Cons truction of Target Populations. American Political Science Review 87 (2), pp. 334 346 Schneider, A. & Ingram, H. ( 1997 ) Policy Design for Democracy Lawrence, KS: University Press of Kansas. Schneider, M. & Teske P ( 1993 ). The Antigrowth Entrepreneur: Challenging the Equi librium of the Growth Machine. The Journal of Politics 55 (3), pp. 720 736. Schewenius, M., McPhearson, T., & Eimqvist, T. (2014). Opportunities for Increasing Resiliance and Sustainability of Urban Social Ecological Systems: Insights from the URBES and the Cities and Biodiversity Outlook Projects. AMBIO: A Journal of the Human Environment (43), pp. 434 444. Schlager, E. & Weible, C. M. (2013). New Theories of the Policy Process. Policy Studies Journal, 41 (3), pp. 389 396. Schnoor, J. ( 2005 ). Global Warming: A Consequence of Human Activities Rivaling Ea rth's Biogeochemical Processes. Human and Ecological Risk Assessment 11 (6), pp. 1105 1110. Selin, H. S. & VanDeveer, S. D. ( 2007 ). Political Science an d Prediction: What's Next for U.S. Climate Change Policy. Review of Policy Research 24 (1), pp. 1 27. Sewell, G. C. ( 2005 ). Actors, Coalitions, and the Framework Convention on Climate Change Unpublished Dissertation: Massach usetts Institute of Technology. Shanahan, E. A., Jones, M. D., & McBeth, M. K. (2011). Policy Narratives and Policy Process. Policy Studies Journal 39 (3), pp. 535 561.

PAGE 193

! 179 Ship an, C. R. & Volden, C. ( 2006 ) Bottom Up Federalism: The Diffusion of Antismoking Polic ies from U.S. Cities to States. American Journal of Political Science 50 (4), pp. 825 843. Shipan, C. R. & Volden, C. (2008). The Mechanics of Policy Diffusion. American Journal of Political Science, 52 : pp. 840 857. Shoemaker, P. J., Tankard, Jr., J. W. & Lasora, D. L. ( 2004 ) How to Build Social Science Theories Thousand Oa ks, CA: Sage Publications, Inc. Siegenthaler, U., Stocker, T. F., Monnin, E., Luthi, D., Schwander, J., Stauffer, B., Raynaud, D., Barnola, J. M., Fi scher, H., Masson Delmotte, V. & Jouzel, J. ( 2005 ). Stable Carbon Cycle Cl imate Relationship During the Late Pleistocene. Science 310 (5752), pp. 1313 1317 Simon, H. A. ( 1985 ). Human nature in politics: The dialogue of ps ychology with political science. Am erican Political Science Review, 79 (2), pp. 293 304. Stern, K. ( 2007 ) The Economic of Climate Change: The Stern Review London, UK: Cambridge University Press Stern, P. & Dietz, T. (2015). IPCC: Social Scientists are Ready. Nature, 521 p. 161. Teske, P. ( 1992 ). Toward a Theory of the Political Entrepreneur: Evidence from Local Government. American Political Science Review 86 (3), pp. 737 747. U S M ayors Conference of Mayors. ( 2015 ) Climate Protection Agreement. http://www.usmayors.org/climateprotection/agreement.htm Last accessed 4/29/15. Vig, N. J. & Kraft, M. E. ( 2016 ) Environmental Policy: New Directions for the Twenty First Century, (9 th ed.) Washington D.C.: CQ Press. 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 Research and Theory 17 (1), pp. 95 117 Weible, C. M. ( 2008 ). Expert Based Information and Policy Subsystems: A Review and Synthesis. The Policy Studies Journal, 36 (4), pp. 615 6 35 Weible, C. M. ( 2014a ) "Advancing Policy Process Research." I n P. A. Sabatier and C. M. Weible (Ed s .) Theories of the Policy Process (3 rd ed.) Boulder, CO: Westview Press. Weible, C. M. ( 2014 b ) "Introducing the Scope and Focus of Policy Process Research and Theory." I n P. A. Sabatier and C. M. Weible (Ed s .) Theories of the Policy Process (3 rd ed.) Boulder, CO: Westview Press

PAGE 194

! 180 Weible, C. M. & Elgin, D. E. ( 2013 ). Contrasting Capacities From C ity to International Levels of Government in Addres sing Climate and Energy Issues. Cityscape: A Journal of Policy Development and Research 15 (1), pp. 171 188. Weible, C. M., Elgin, D. E. & Pattison A ( 2012 ). Policy Analytical Capacity and Policy Activities. Canadian Political Science Review 6 (1), pp. 125 137. Weible, C., Pattison, A. & Sabatier P ( 2010 ). Harnessing Expert Based Information for Learning and The Sustainable Management of Co mplex Socio ecological Systems. Environmental Science & Policy 13 pp. 522 534. Weible, C., Sabatier, P. & McQueen K. ( 2009 ). T hemes and Variations: Taking Stock of the Advocacy Coalition Framework. The Policy Studies Journal 37 (1), pp. 121 140 Weible, C. & Sabatier, P. ( 2008 ). Coalitions, Science, and Belief Change: Comparing Adversarial and C ollaborative Policy Subsystems. The Policy Studies Journal 37 (2), pp. 195 212. Weible, C., Sabatier, P., Jenkins Smith, H. C. Nohrstedt, D., Henry, A. D ., & de Leon, P. ( 2011 ). A Quarter Century of the Advocacy Coalition Framework: An In troduction to the Special Issue Policy Studies Journal 39 (3), pp. 349 360. Weible, C., Sabatier, P. & Lea ch, W. ( 2007 ) "Advocacy Coalitions and Policy Networks: Allies, Opponents, and Information in Watershed Partnerships. In An Advocacy Coalition Lens on Environmental Policy U nder revi ew at MIT Press, Cambridge, MA. Wood, R. S., Hultquist, A., & Romsdahl, R. J. ( 2014 ). An Examination of Local Clima te Policies in the Great Plains Review of Policy Research 31 (6), pp. 529 554. Zafonte, M. & Sabatier P ( 1998 ). Shared Beliefs a nd Imposed Interdependence as Determinants of Ally Networks in Overlapping Subsystems. Journal of Theoretical Politics 10 (4), pp. 473 505. Zahariadis, N. (2014). "Ambiguity and Multiple Streams". I n P. A. Sabatier and C. M. Weible (Ed s .) Theories of the Policy Process (3 rd ed.) Boulder, CO: Westview Press. Zahra n, S., Grover, H., Brody, S. D. & Vedlitz A ( 2008 ). Risk, Stress, and Capacity: Explaining Metropolitan C ommitment to Climate Protection Urban Affairs Review 43 (4), pp. 447 474. Zaller, J. R. ( 1992 ) The nature and origins of mass o pinion Cambridge, UK: Cambri dge University Press

PAGE 195

! 181 APPENDIX A Output from Chapter Three Multinomial Logit Regression Learning Regarding the Predicted Severity of the Impacts of Climate Change Number of Observations = 251. Chi 2 ( 16 ) = 1 10 64 Prob > Chi 2 = 0.0000. Pseudo R 2 = 0. 17 Baseli ne is Belief Reinforcement Only Coef. Std. err. Z. P value 95% Conf. I nterval _________________________________________________________________________ _____ 1: Belief Reinforcement Only (base outcome) _______________________________________________________________________________ 2: My Views Have Been Mostly Reinforcement with Little Changes Extr emeness of Belief 1.8 2 0.40 4. 52 0.00 2. 60 1.0 3 Freq. of Advice with Agree 0.70 0.21 3.36 0.00 1.10 0.28 Freq. of Collab orate with Disagree 0.31 0.17 1.87 0.06 0.02 0.64 Freq. of Facilitated Consensus 0.13 0.17 0.75 0.46 0.46 0.21 Constant 4. 6 9 0.9 6 4. 87 0.00 2. 80 6.5 9 _______________________________________________________________________________ 3: My Views Have had a Balance of Reinforcement and Changes Extremeness of Belief 2.81 0.43 6. 61 0.00 3.6 4 1.97 Freq. of Advice with Agree 0.4 1 0.25 1. 69 0.09 0.9 0 0.06 Freq. of Collaborate with Disagree 0.0 2 0.20 0. 11 0.9 1 0.3 7 0.4 1 Freq. of Facilitated Consensus 0.1 5 0.20 0. 73 0. 46 0.2 4 0.5 3 Constant 5 02 1.0 3 4. 86 0.00 3.00 7 05 _______________________________________________________________________________ 4 : My Views Have Mostly Changed with Little Reinforcement Extremeness of Belief 1. 91 1. 41 1.35 0.1 7 4 69 0. 86 Freq. of Advice with Agree 0. 70 1. 20 0. 58 0. 56 3 05 1 66 Freq. of Collaborate with Disagree 14 02 542 9 5 0.03 0.98 1050 13 1078 18 Freq. of Facilitated Consensus 0.79 0 77 1 .0 3 0. 30 0.71 2.29 Constant 53 .90 2171 79 0.0 2 0.98 4310 52 4202.75 _______________________________________________________________________________ 6 : My Views Have Neither Changed nor Been Reinforced Extremeness of Belief 3. 06 0.5 2 5.8 3 0.00 4. 08 2.0 3 Freq. of Advice with Agree 0. 31 0.3 7 0. 85 0. 40 1 03 0. 41 Fre q. of Collaborate with Disagree 0.04 0.3 0 0.1 2 0.9 1 0. 63 0. 56 Freq. of Facilitated Consensus 0. 27 0. 29 0. 94 0. 35 0. 29 0. 84 Constant 3.64 1.3 0 2 80 0.00 1. 09 6. 20 _______________________________________________________________________________

PAGE 196

! 182 Learning Regarding Humans as the Cause of Climate Change Number of Observations = 249. Chi 2 (2 0 ) = 1 05.70 Prob > Chi 2 = 0.0000. Pseudo R 2 = 0. 16 Baseline is Belief Reinforcement Only Coef Std err. Z. P value 95% Conf. I nterval _______________________________________________________________________________ 1: Belief Reinforcement Only (base outcome) _______________________________________________________________________________ 2: My Views Have Been Mostly Reinforcement with Little Changes Extr emeness of Belief 1. 46 0.3 3 4. 3 8 0.00 2. 12 0.8 1 Freq. of Advice with Agree 0.2 0 0.19 1. 06 0.2 9 0. 57 0.1 7 Freq. of Collaborate with Disagree 0.0 5 0.16 0. 33 0. 74 0.2 6 0.3 6 Freq. of Facilitated Consensus 0.0 4 0.17 0. 28 0. 78 0.3 7 0. 28 Constant 2. 65 0.80 3. 33 0.00 1.09 4. 20 _______________________________________________________________________________ 3: My Views Have had a Balance of Reinforcement and Changes Extremeness of Belief 3.1 1 0.4 2 7.3 5 0.00 3 94 2. 28 Freq. of Advice with Agree 0.0 5 0.25 0. 19 0. 85 0.5 3 0.4 4 Freq. of Collaborate with Disagree 0.17 0.21 0.8 0 0.4 3 0.59 0.25 Freq. of Facilitated Consensus 0. 07 0.2 1 0. 35 0. 73 0.3 4 0. 49 Constant 4 05 0.9 2 4. 41 0.00 2. 25 5 .86 _______________________________________________________________________________ 4: My Views Have Mostly Changed with Little Reinforcement Extremeness of Belief 3. 66 2. 92 1 25 0. 21 9. 38 2 07 Freq. of Advice with Agree 1. 37 1 .55 0. 88 0. 38 4 41 1 68 Fre q. of Collaborate with Disagree 18 87 5017 40 0.0 0 1 00 9815 05 9852 79 Freq. of Facilitated Consensus 0 .1 9 1.11 0.1 8 0. 86 1.98 2. 39 Constant 68 97 20069.59 0.0 0 1 00 3 9404.64 3 9266 71 _______________________________________________________________________________ 5: My Views have Changed Only Extremeness of Belief 3 2 18 2483 .6 6 0.0 1 0.99 4900 07 4835 71 Freq. of Advice with Agree 0 75 2432 .0 4 0.00 1.00 4765 96 4767 45 Freq. of Collaborate with Disagree 9 76 2420 12 0.00 1.00 4733 69 4753.12 Freq. of Facilitated Consensus 10. 56 1840 52 0.0 1 1.00 3596 79 3617 .9 2 Constant 6 5 31 7361 50 0.0 1 0 99 14493 .5 8 14362 96 _______________________________________________________________________________ 6: My Views Have Neither Changed nor Been Reinforced Extremeness of Belief 1. 58 0.4 5 3. 51 0.00 2. 46 0.6 9 Freq. of Advice with Agree 0. 12 0.28 0. 44 0. 66 0.6 8 0.4 3 Freq. of Collaborate with Disagree 0.02 0.24 0. 10 0.9 2 0.4 4 0.49 Freq. of Facilitated Consensus 0.2 0 0.23 0. 86 0. 39 0. 26 0. 65 Constant 1. 14 1. 08 1. 06 0.2 9 0. 98 3. 2 8 _______________________________________________________________________________

PAGE 197

! 183 Learning Regarding the Need for Government to Address Climate Change Number of Observations = 249. Chi 2 (2 0 ) = 61.18 Prob > Chi 2 = 0.0000. Pseudo R 2 = 0. 08 Baseline is Belief Reinforcement Only Coef. Std err. Z. P value 95% Conf. Interval _______________________________________________________________________________ 1: Belief Reinforcement Only (base outcome) _______________________________________________________________________________ 2: My Views Have Been Mostly Reinforcement with Litt le Changes Extremeness of Belief 1.47 0.39 3.7 6 0.00 2.24 0.70 Freq. of Advice with Agree 0. 10 0.22 0.4 5 0.6 6 0.5 0 0.3 3 Freq. of Collaborate with Disagree 0.06 0.18 0.3 4 0.7 4 0.30 0.42 Freq. of Facilitated Consensus 0.26 0.20 1.33 0.18 0.64 0.12 Constant 2. 96 0. 88 3. 37 0.00 1. 24 4. 68 _______________________________________________________________________________ 3: My Views Have had a Balance of Reinforcement and Changes Extremeness of Belief 2.0 9 0.39 5.3 8 0.00 2.8 6 1.3 3 Freq. of Advice with Agree 0.1 2 0.23 0.5 1 0. 61 0.5 6 0.3 3 Freq. of Collaborate with Disagree 0.16 0.19 0.8 7 0. 38 0.2 0 0.53 Freq. of Facilitated Consensus 0.2 1 0. 19 1. 1 1 0. 27 0.59 0.1 6 Constant 3. 62 0. 87 4 14 0.00 1.9 0 5. 34 _______________________________________________________________________________ 4: My Views Have Mostly Changed with Little Reinforcement Extremeness of Belief 1.27 0.9 4 1.3 5 0.18 3.1 0 0.5 7 Freq. of Advice with Agree 0.5 7 0.5 9 0.9 7 0.3 3 1. 72 0. 58 Freq. of Collaborate with Disagree 1. 57 0.6 5 2. 39 0.02 0.2 8 2. 85 Freq. of Facilitated Consensus 0.5 8 0.5 5 1. 07 0.2 9 1.65 0.4 9 Constant 2. 31 2. 58 0. 90 0.3 7 7. 39 2. 75 _______________________________________________________________________________ 5: My Views have Changed Only Extremeness of Belief 1.3 2 0.90 1.4 6 0.14 3.0 8 0.4 5 Freq. of Advice with Agree 0.0 9 0.68 0.1 3 0.90 1.2 5 1.43 Freq. of Collaborate with Disagree 0.9 6 0. 59 1.6 3 0.10 0.19 2.1 0 Freq. of Facilitated Consensus 0.2 3 0. 49 0. 46 0.6 5 1. 19 0.7 4 Constant 2.8 4 2.6 4 1.0 8 0. 28 8 01 2. 32 _______________________________________________________________________________ 6: My Views Have Neither Changed nor Been Reinforced Extremeness of Belief 1.8 1 0.4 0 4.49 0.00 2.6 0 1.0 2 Freq. of Advice with Agree 0.0 2 0.2 4 0. 08 0. 93 0.4 6 0. 50 Freq. of Collaborate with Disagree 0.2 0 0.2 0 1. 02 0. 31 0.1 9 0. 59 Freq. of Facilitated Consensus 0. 07 0.20 0. 36 0. 72 0. 46 0. 32 Constant 2. 25 0.9 3 2. 41 0.0 2 0. 42 4. 09 _______________________________________________________________________________

PAGE 198

! 184 Learning Regarding the Need for a Carbon Tax to Address Climate Change Number of Observations = 249. Chi 2 (2 0 ) = 42.76 Prob > Chi 2 = 0.00 22 Pseudo R 2 = 0. 06 Baseline is Belief Reinforcement Only Coef. Std err. Z. P value 95% Conf. Interval _______________________________________________________________________________ 1: Belief Reinforcement Only (base outcome) _______________________________________________________________________________ 2: My Views Have Been Mostly Reinforcement with Little Changes Extremeness of Belief 0. 55 0. 29 1 88 0.0 6 0.0 2 1. 12 Freq. of Advice with Agree 0.0 5 0.23 0. 22 0.8 3 0. 39 0. 49 Freq. of Collaborate with Disagr ee 0.1 3 0.19 0. 67 0. 50 0. 49 0.2 4 Freq. of Facilitated Consensus 0. 15 0. 20 0 79 0. 43 0. 23 0.5 4 Constant 0. 23 0.7 3 0 32 0. 75 1 67 1 20 _______________________________________________________________________________ 3: My Views Have had a Balance of Reinforcement and Changes Extremeness of Belief 0. 18 0. 29 0. 6 3 0. 53 0. 75 0. 39 Freq. of Advice with Agree 0.06 0.2 4 0.2 7 0. 79 0.4 0 0.5 3 Freq. of Collaborate with Disagree 0.0 5 0.2 0 0. 2 3 0.8 2 0.3 4 0.43 Freq. of Facilitated Consensus 0. 15 0.2 0 0 74 0. 46 0.2 5 0. 55 Constant 0. 11 0. 74 0.1 5 0. 88 1 34 1 57 _______________________________________________________________________________ 4: My Views Have Mostly Changed with Little Reinforcement Extremeness of Belief 0. 74 0. 69 1. 07 0.2 8 0. 61 2.0 9 Freq. of Advice with Agree 0.1 7 0.51 0. 33 0. 74 0. 83 1.1 7 Freq. of Collaborate with Disagree 0. 43 0.4 0 1 07 0. 29 0. 36 1. 22 Freq. of Facilitated Consensus 0. 40 0.4 3 0. 92 0. 36 1.2 4 0.4 5 Constant 3 79 1.8 2 2. 08 0.0 4 7 35 0. 22 _______________________________________________________________________________ 5: My Views have Changed Only Extremeness of Belief 2. 08 1. 14 1.8 2 0.07 4. 31 0.1 5 Freq. of Advice with Agree 0.3 7 0.6 7 0. 56 0.5 8 1. 6 8 0 9 4 Freq. of Collaborate with Disagree 0. 65 0. 66 0 99 0. 32 0.6 3 1 94 Freq. of Facilitated Consensus 0.22 0.6 0 0.3 6 0.7 2 0.9 6 1.4 0 Constant 1 89 2. 04 0 92 0. 36 5 .8 9 2 12 _______________________________________________________________________________ 6: My Views Have Neither Changed nor Been Reinforced Extremeness of Belief 0. 72 0.3 1 2. 35 0.0 2 1. 33 0. 12 Freq. of Advice with Agree 0.1 2 0.25 0. 49 0. 63 0.6 1 0.3 7 Fre q. of Collaborate with Disagree 0.3 9 0.2 1 1. 83 0.0 7 0.0 3 0. 81 Freq. of Facilitated Consensus 0.0 6 0.22 0.2 8 0. 78 0.3 6 0.48 Constant 0. 36 0.7 6 0. 47 0. 64 1 13 1 .8 5 _______________________________________________________________________________

PAGE 199

! 185 Learning Regarding the Need for a Cap & Trade System Number of Observations = 246. Chi 2 (2 0 ) = 58.43 Prob > Chi 2 = 0.0000. Pseudo R 2 = 0. 07 Baseline is Belief Reinforcement Only Coef. Std err. Z. P value 95% Conf. Interval _______________________________________________________________________________ 1: Belief Reinforcement Only (base outcome) _______________________________________________________________________________ 2: My Views Have Been Mostly Reinforcement with Little Changes Extremeness of Belief 0.6 4 0.3 6 1. 77 0.0 8 1.3 5 0.0 7 Freq. of Advice with Agree 0.04 0.26 0. 0 1 0. 99 0.5 0 0. 50 Freq. of Collaborate with Disagree 0.0 1 0.21 0.0 3 0.9 8 0.4 0 0.41 Freq. of Facilitated Consensus 0. 20 0.2 1 0.9 4 0. 3 5 0. 62 0. 22 Constant 1. 69 0.8 6 1. 96 0.0 5 0. 00 3. 38 _______________________________________________________________________________ 3: My Views Have had a Balance of Reinforcement and Changes Extremeness of Belief 1. 6 3 0.3 6 4. 49 0.00 2. 3 4 0.9 2 Freq. of Advice with Agree 0.2 9 0.2 6 1 0 9 0. 28 0.2 3 0.81 Freq. of Collaborate with Disagree 0.0 5 0.2 1 0. 2 3 0.8 2 0. 37 0.4 6 Freq. of Facilitated Consensus 0. 19 0.22 0. 88 0. 38 0. 61 0. 23 Constant 2 .0 3 0. 88 2 33 0.0 2 0. 33 3. 73 _______________________________________________________________________________ 4: My Views Have Mostly Changed with Little Reinforcement Extremeness of Belief 0.9 1 0.53 1.7 3 0.08 1.9 5 0.1 2 Freq. of Advice with Agree 0.2 6 0.42 0.6 1 0.5 4 0.5 7 1.0 8 Freq. of Collaborate with Disagree 0.17 0.33 0.5 2 0.6 1 0.81 0.47 Freq. of Facilitated Consensus 0.3 1 0.3 2 0 95 0.3 4 0.3 3 0.9 4 Constant 0. 79 1. 39 0. 57 0.5 7 3. 50 1. 93 _______________________________________________________________________________ 5: My Views have Changed Only Extremeness of Belief 2.14 0.7 5 2.8 5 0.00 3.6 1 0.6 7 Freq. of Advice with Agree 0.3 7 0.58 0.6 4 0.5 2 0.7 7 1. 51 Freq. of Collaborate with Disagree 0. 60 0.48 1.2 4 0.22 1.54 0.3 5 Freq. of Facilitated Consensus 0. 49 0.48 1.0 4 0. 30 0.4 4 1.4 3 Constant 0.3 1 1.7 5 0.1 7 0.8 6 3. 74 3.1 3 _______________________________________________________________________________ 6: My Views Have Neither Changed nor Been Reinforced Extremeness of Belief 1. 83 0.37 4. 95 0.00 2. 56 1. 11 Freq. of Advice with Agree 0.0 6 0.27 0.2 4 0. 81 0. 59 0.4 6 Freq. of Collaborate with Disagr ee 0.3 0 0.22 1. 3 5 0.1 8 0.1 3 0.7 3 Freq. of Facilitated Consensus 0.2 2 0.22 1. 0 1 0.3 1 0. 66 0. 2 1 Constant 2. 68 0.8 7 3. 10 0.00 0 98 4. 37 _______________________________________________________________________________

PAGE 200

! 186 Learning Regarding the Need for Policies Promoting Renewables Number of Observations = 247. Chi 2 (2 0 ) = 1 09 85 Prob > Chi 2 = 0.0000. Pseudo R 2 = 0. 16 Baseline is Belief Reinforcement Only Coef. Std err. Z. P value 95% Conf. Interval _______________________________________________________________________________ 1: Belief Reinforcement Only (base outcome) _______________________________________________________________________________ 2: My Views Have Been Mostly Reinforcement with Litt le Changes Extremeness of Belief 3.2 5 0.53 6.1 8 0.00 4. 28 2.2 2 Freq. of Advice with Agree 0.1 3 0.22 0.5 8 0. 5 6 0.55 0.3 0 Freq. of Collaborate with Disagree 0.2 7 0.19 1. 43 0.1 5 0.6 3 0. 10 Freq. of Facilitated Consensus 0.05 0.19 0.2 8 0. 78 0.3 2 0.43 Constant 5. 81 1.1 3 5. 12 0.00 3. 59 8.0 3 _______________________________________________________________________________ 3: My Views Have had a Balance of Reinforcement and Changes Extremeness of Belief 3.12 0.5 6 5. 76 0.00 4.3 0 2.01 Freq. of Advice with Agree 0.2 2 0.2 6 0.8 4 0.40 0.30 0.7 4 Freq. of Collaborate with Disagree 0.1 8 0.2 1 0. 84 0. 4 0 0.2 4 0.5 9 Freq. of Facilitated Consensus 0.16 0.22 0.7 4 0.4 6 0. 58 0.2 6 Constant 3. 69 1.2 4 2. 97 0.0 0 1 26 6 13 _______________________________________________________________________________ 4: My Views Have Mostly Changed with Little Reinforcement Extremeness of Belief 3.3 2 1.1 3 2. 94 0.0 0 5. 54 1. 11 Freq. of Advice with Agree 0. 79 0.8 0 0 99 0. 32 0.7 8 2. 36 Freq. of Collaborate with Disagree 0. 21 0.57 0. 36 0.7 2 0.9 1 1. 33 Freq. of Facilitated Consensus 0. 77 0. 68 1.1 3 0.2 6 2. 09 0.5 6 Constant 0.28 3. 03 0.09 0.93 5 .6 6 6 22 _______________________________________________________________________________ 5: My Views have Changed Only Extremeness of Belief 3.3 2 1.1 0 3 02 0.0 0 5.4 8 1 16 Freq. of Advice with Agree 0.07 0.74 0.10 0.92 1.5 3 1.3 8 Freq. of Collaborate with Disagree 0.5 4 0.6 2 0. 86 0. 39 0. 69 1.7 6 Freq. of Facilitated Consensus 0.0 4 0.58 0.0 7 0.9 4 1.1 7 1. 09 Constant 1 24 2 77 0. 45 0. 66 4 .1 8 6. 66 _______________________________________________________________________________ 6: My Views Have Neither Changed nor Been Reinforced Extremeness of Belief 3. 53 0.5 4 6. 5 7 0.00 4. 58 2. 4 7 Freq. of Advice with Agree 0.05 0.2 4 0.1 9 0.8 5 0.5 2 0.4 2 Freq. of Collaborate with Disagree 0.2 0 0. 20 1. 01 0. 31 0.1 9 0. 59 Freq. of Facilitated Consensus 0.1 4 0.2 0 0.7 1 0.4 5 0.5 4 0.2 6 Constant 5 .1 5 1. 17 4. 4 2 0.00 2 8 6 7 .4 3 _______________________________________________________________________________

PAGE 201

! 187 APPENDIX B Output from Chapter Four Multinomial Logit Regression Learning Regarding Advocacy Strategy for Cl imate and Energy Policy Issues Number of O bs ervations = 2 51 Chi 2 (15) = 40 55 Prob > C hi 2 = 0.000 4 Pseudo R 2 = 0. 05 Baseline is My Views Have Only Been Reinforcement Coef. Std err. z P >|z| 95% Conf. Interval _______________________________________________________________________________ 1: My Views Have Only Been Reinforcement (base outcome) _______________________________________________________________________________ 2: My Views Have Been Mostly Reinforcement with Little Changes Extremeness of Policy Beliefs 0.2 1 0. 09 2. 33 0.0 2 0. 39 0.0 3 Coalition Building 0. 23 0. 39 0 59 0. 55 0. 53 1 00 Consensus Based Processes 0. 59 0. 39 1. 5 1 0. 13 1. 35 0. 18 Constant 2 .0 7 0 88 2 37 0.0 2 0 36 3 79 _______________________________________________________________________________ 3: My Views Have had a Balance of Reinforcement and Changes Extremeness of Policy Beliefs 0.1 9 0.09 2 20 0.0 3 0.3 7 0.0 2 Coalition Building 0. 33 0. 37 0. 90 0. 37 0.3 9 1 06 Consensus Based Processes 0. 04 0.37 0.1 0 0. 92 0. 77 0. 6 9 Constant 1. 78 0. 85 2 10 0.0 4 0. 12 3. 44 _______________________________________________________________________________ 4: My Views Have Mostly Changed with Little Reinforcement Extremeness of Policy Beliefs 0. 17 0.1 6 1. 06 0. 29 0. 49 0. 15 Coalition Building 0. 25 0. 7 1 0. 35 0. 73 1 15 1 .6 5 Consensus Based Processes 1. 42 0 76 1. 88 0.0 6 2 9 0 0. 06 Constant 0. 25 1. 49 0. 1 7 0. 87 2. 67 3 17 _______________________________________________________________________________ 5: My Views have Changed Only Extremeness of Policy Beliefs 0.2 2 0.2 4 0. 91 0. 36 0. 67 0. 25 Coalition Building 0 82 1 20 0 .68 0. 50 1 53 3 17 Consensus Based Processes 0. 63 1. 06 0. 60 0. 55 2. 71 1. 44 Constant 0 95 2 23 0 43 0. 67 5 34 3 41 _______________________________________________________________________________ 6: My Views Have Neither Changed nor Been Reinforced Extremeness of Policy Beliefs 0. 51 0.1 0 5. 02 0.00 0. 71 0.3 1 Coalition Building 0. 48 0. 49 1 04 0. 30 0. 42 1 38 Consensus Based Processes 0 49 0 44 1 11 0. 27 1 36 0 38 Constant 3 78 0 9 0 4. 19 0.00 2 01 5 55 _______________________________________________________________________________