Mobilizing After Disasters in Advanced Industrial Democracies

MOBILIZING AFTER DISASTERS IN ADVANCED INDUSTRIAL DEMOCRACIES

MIRIAM MATEJOVA

B.A. (Honours), University of Northern British Columbia, 2009 M.A., Norman Paterson School of International Affairs, Carleton University, 2012

A DISSERTATION SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF

DOCTOR OF PHILOSOPHY

THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES

(Political Science)

THE UNIVERSITY OF BRITISH COLUMBIA

(Vancouver)

April 2019

The following individuals certify that they have read, and recommend to the Faculty of Graduate and Postdoctoral Studies for acceptance, the dissertation entitled:

Mobilizing after disasters in advanced industrial democracies

in partial fulfillment of the requirements for submitted by Miriam Matejova the degree of Doctor of Philosophy in Political Science

Examining Committee: Peter Dauvergne

Supervisor Lisa McIntosh Sundstrom

Supervisory Committee Member Catherine Corrigall-Brown

Supervisory Committee Member Brian Job

University Examiner Terre Satterfield

University Examiner

Additional Supervisory Committee Members:

Supervisory Committee Member

Abstract

Environmental disasters are frequently catalysts for social and political change. Yet, disasters of similar scale and impact seem to encourage collective action in some cases but fail to do so in others. For example, while some large oil spills have generated mass nationwide (and international) protests, others have gone largely unnoticed and protests, if any, remained small and localized. If disasters are political triggering events, as the existing literature suggests, why do they often fail to generate large scale collective action? In fact, why do some highly damaging industrial environmental disasters succeed, and others fail to catalyze protest movements? This research strives to explain a variation in the occurrence and size of non-violent protest after industrial environmental disasters in advanced democracies. I examine the mobilizing effects of disaster type and location, the underlying societal conditions conducive to protest, and the ‘language of disasters’ in post-disaster communication. I argue that in addition to grievances, resources, political opportunities, and framing, uncertainty about disaster impacts plays a crucial role in the protest mobilization process, one that has not been fully explored by scholars. Specifically, while uncertainty may have a dampening effect on protest mobilization, this effect is conditioned by people’s pre-existing beliefs, and particularly political ideology. Left-leaning (i.e., more liberal) individuals resist the dampening effect of uncertainty, while right-leaning (i.e., more conservative) individuals embrace it. This research draws on theories of social movements and framing as well as insights from previously studied disasters; it involves an in-depth analysis of cases of industrial disasters with large environmental impacts, including the 2014 Mount Polley mine leak, the 2010 Deepwater Horizon oil spill, and the 2011 Fukushima nuclear disaster. The cases were selected due to the varying protest sizes following these events. To allow for a systematic examination of different factors linked to post-disaster protest, this research employs several methods and tools, including a geographic information systems (GIS) analysis, qualitative comparative analysis (QCA), content analysis, and survey experiment. Such multi-method approach is most suitable for answering the variety and complexity of questions this research poses.

iii

Lay Summary

This dissertation examines industrial environmental disasters – oil spills, mine leaks, and nuclear accidents – and their effect on nonviolent protest mobilization in advanced industrial democracies. Surprisingly, large environmental disasters are often met with little public response. Post-disaster protests – if they emerge at all – tend to be small and localized. To some extent, grievances, the underlying socio-economic conditions, and framing activities explain the variation in post-disaster protest emergence and growth. An additional element – one scholars have largely neglected – is uncertainty about disaster impacts, which shapes public willingness to protest through individuals’ pre-existing beliefs. Specifically, among more politically liberal individuals, uncertainty fails to discourage protest. The effect is the opposite among more conservative individuals. This comprehensive examination of industrial environmental disasters enhances our understanding of the protest mobilization process and disasters’ social impacts. Some findings can be used to improve disaster communication practices, and open opportunities for peaceful resolution of social conflict.

Preface

This dissertation is the original, unpublished, independent work of the author, Miriam Matejova. The research (and specifically Chapter 6) was approved by the University of British Columbia’s (UBC) Behavioral Research Ethics Board (BREB), BREB Number H18-00999. The experiment in Chapter 6 was conducted in collaboration with Eric Merkley. The geospatial analysis in Chapter 3 was conducted in collaboration with Devin Lussier.

Table of Contents Abstract ...... iii Lay Summary ...... iv Preface ...... v Table of Contents …………….………………………………………………………………………………………………….………………… vi List of Tables ...... viii List of Figures ...... x Acknowledgements ...... xi Dedication ………………………………………………………………………………………………………………….… xiii Chapter 1. Introduction ...... 1 Chapter 2. Disasters and Protest: Definitions and Theoretical Framework ...... 9 1. Disasters in Social Science ...... 9 1.1 Disaster typologies ...... 12 1.2 Industrial environmental disasters ...... 15 2. Social Movements and Post-Disaster Protests ...... 17 2.1 Grievances, resources, and political opportunity structures ...... 21 2.2 Framing and post-disaster protest mobilization ...... 24 Conclusion ...... 26 Chapter 3. Disaster Damage and Sudden Grievances: Mapping Environmental Disasters and Protest ..... 27 1. Environmental Disasters and Grievances ...... 28 1.1 Grievances and environmental values ...... 29 1.2 Grievances and disaster proximity ...... 31 1.3 Hypotheses ...... 33 2. GIS Analysis ...... 36 2.1 GIS, social science, and disasters ...... 36 2.2 Data, analysis, and mapping ...... 37 2.3 Discussion ...... 43 Conclusion ...... 46 Chapter 4. Structural Conditions for Post-Disaster Protest ...... 52 1. Structural Conditions, Disasters, and Protest ...... 53 2. Examining Post-Disaster Protest through QCA ...... 54 2.1 Theory and raw data ...... 55 2.2 Set calibration and coding ...... 58 2.3 Analysis and discussion ...... 64

Conclusion ...... 73 Chapter 5. Disaster Language ...... 75 1. Frames and the Framing Research ...... 76 2. Disaster Framing and Post-Disaster Uncertainty ...... 77 2.1 Framing actors ...... 80 2.2 Types of post-disaster frames ...... 82 2.3 Expectations ...... 87 3. Content Analysis ...... 88 3.1 Case selection and cases ...... 89 3.2 Data sources and coding...... 91 3.3 Results and discussion ...... 94 Conclusion ...... 105 Chapter 6. Uncertainty Framing of Environmental Disasters and the Willingness to Protest ...... 106 1. The Framing Theory Continued ...... 107 1.1 Frame effectiveness and frame strength ...... 108 1.2 Negative frames, Prospect Theory, and post-disaster protest ...... 109 2. Towards a Better Understanding of Protest: Uncertainty and the Willingness to Protest ...... 111 3. Experiment ...... 113 3.1 Design and methods ...... 114 3.2 Data...... 117 3.3. Results ...... 120 Conclusion ...... 124 Chapter 7. Conclusion ...... 126 Bibliography ...... 133 Appendix 1 ...... 175 Appendix 2 ...... 183 Appendix 3 ...... 189 Appendix 4 ...... 198 Appendix 5 ...... 203 Appendix 6 ...... 213

vii

List of Tables

Table 1. Typology of environmental disasters according to their source and onset...... 13 Table 2. Large oil spills in advanced industrial democracies, 1900 – present...... 18 Table 3. Large mine leaks in advanced industrial democracies, 1900 – present...... 20 Table 4. Large nuclear disasters in advanced industrial democracies, 1900 – present...... 21 Table 5. Environmental values and pollution from industrial disasters...... 30 Table 6. Coding post-disaster protest data...... 38 Table 7. Measuring social movement strength...... 57 Table 8. QCA conditions...... 59 Table 9. Concepts, indicators, and set membership thresholds...... 63 Table 10. Analysis of necessary conditions for the outcome ‘protest’...... 65 Table 11. Truth table for the outcome ‘protest’...... 67 Table 12. Conservative solution for the outcome ‘protest’...... 68 Table 13. Analysis of necessary conditions for the outcome ‘absence of protest’...... 69 Table 14. Truth table for the outcome ‘absence of protest’...... 72 Table 15. Conservative solution for the outcome ‘absence of protest’...... 73 Table 16. Post-disaster frames in news coverage...... 84 Table 17. Tone of post-disaster frames...... 86 Table 18. The prevailing topics in the news coverage after the Mount Polley, Deepwater Horizon, and Fukushima disasters...... 94 Table 19. Incidence of frames in disaster news coverage (in percent, rounded)...... 95 Table 20. Percentage (rounded) of each framing variable...... 97 Table 21. Tone of news coverage (incidence in percent, rounded)...... 100 Table 22. Percentage (rounded) of framing actors covered in the post-disaster news...... 103 Table 23. Percentage (rounded) of frames out of all statements framing actors made in the news...... 103 Table 24. Uncertainty in the news media coverage of the Mount Polley, Deepwater Horizon, and Fukushima disasters (in percent)...... 112 Table 25. Treatment Conditions...... 115 Table 26. Comparison of 2016 GSS survey and Amazon Mechanical Turk sample...... 119 Table 27. Difference of Means Tests, H1-H3...... 121 Table 28. OLS Regression Estimates, H4...... 122 Table 29. Robustness tests, OLS estimates...... 124 Table 30. Disaster locations and areas impacted...... 175 Table 31. Post-disaster protest events...... 180 Table 32. Analysis of necessary conditions for the outcome ‘small protest’...... 183 Table 33. Analysis of necessary conditions for the outcome ‘medium protest’...... 183 Table 34. Analysis of necessary conditions for the outcome ‘large protest’...... 184 Table 35. Truth table for the outcome ‘small protest’...... 185 Table 36. Truth table for the outcome ‘medium protest’...... 186 Table 37. Truth table for the outcome ‘large protest’...... 187 Table 38. Conservative solution for the outcome ‘medium protest’...... 188

viii

Table 39. Conservative solution for the outcome ‘large protest’...... 188 Table 40. Factor loadings, principal components analysis...... 218

List of Figures

Figure 1. Protest mobilization process...... 24 Figure 2. Environmental disasters and associated protests: Europe, 1900–present...... 40 Figure 3. Environmental disasters and associated protests: North America, 1900–present...... 41 Figure 4. Environmental disasters and associated protests: Asia, 1900–present...... 42 Figure 5. Nuclear power plant facilities and protest before the Fukushima disaster...... 50 Figure 6. Nuclear power plant facilities and protest after the Fukushima disaster...... 51 Figure 7. Example article – no uncertainty frames/no human safety...... 116 Figure 8. Power analysis results at a 0.05 significance level and 0.8 power...... 118 Figure 9. Effect of uncertainty frame on potential protest mobilization conditional on ideology...... 123 Figure 10. LDA topic model for the Mount Polley disaster (10 topics)...... 198 Figure 11. LDA topic model for the Deepwater Horizon spill (10 topics)...... 200 Figure 12. LDA topic model for the Fukushima disaster (Germany) (10 topics)...... 202 Figure 13. Tone and framing actors: journalist, Mount Polley...... 203 Figure 14. Tone and framing actors: activist, Mount Polley...... 203 Figure 15. Tone and framing actors: local government, Mount Polley...... 204 Figure 16. Tone and framing actors: provincial government, Mount Polley...... 204 Figure 17. Tone and framing actors: federal government, Mount Polley...... 205 Figure 18 Tone and framing actors: company, Mount Polley...... 205 Figure 19. Tone and framing actors: expert, Mount Polley...... 206 Figure 20. Tone and framing actors: other, Mount Polley...... 206 Figure 21. Tone and framing actors: journalist, Deepwater Horizon...... 207 Figure 22. Tone and framing actors: activist, Deepwater Horizon...... 207 Figure 23. Tone and framing actors: federal government, Deepwater Horizon...... 208 Figure 24. Tone and framing actors: company, Deepwater Horizon...... 208 Figure 25. Tone and framing actors: expert, Deepwater Horizon...... 209 Figure 26. Tone and framing actors: other, Deepwater Horizon...... 209 Figure 27. Tone and framing actors: journalist, Fukushima...... 210 Figure 28. Tone and framing actors: activist, Fukushima...... 210 Figure 29. Tone and framing actors: federal government, Fukushima...... 211 Figure 30. Tone and framing actors: company, Fukushima...... 211 Figure 31. Tone and framing actors: expert, Fukushima...... 212 Figure 32. Tone and framing actors: other, Fukushima...... 212

Acknowledgements

Pursuing a PhD is a complex, challenging and rewarding journey. Many people have helped me along the way. I am most grateful to my supervisory committee for guidance, feedback, enthusiasm and all the attention they have given to this project. Peter Dauvergne has been a brilliant, inspirational and compassionate supervisor, never wavering in his faith in my abilities. He has taught me how to see puzzles and patterns in world politics, how to find creative answers to complex questions, and how to tell compelling stories about the environment. Peter’s extraordinary attention to detail has taught me the value of careful editing, and his talent for writing inspired me to pay closer attention to my own writing (and especially that dreaded ‘last sentence’). Lisa Sundstrom and Catherine Corrigall-Brown have supported and challenged me, pushing me to carefully think through my assumptions, theories, methods and conclusions. Their enthusiasm for my work has often provided me with much needed reassurance that my research is meaningful. Their feedback and advice have not only contributed greatly to the quality of my work but prompted me to think about who I am (and want to be) as a scholar. In early 2018, I wrote several chapters of this dissertation at the University of Oxford, spending many hours in Oxford’s beautiful libraries. I want to thank the Department of Political Science and International Relations for hosting me. My time in Oxford was one of the most productive, intellectually stimulating and happiest experiences I have ever had. I also want to thank the Purple Dragon School of Martial Arts and Senpai Ioanna Koutava whose kickboxing classes were my ‘happy place’ where I could temporarily let go of theories, methods and literature reviews that occupied most of my waking hours. To Nels Anderson, Paul Dean, Cynthia Hornbeck and Brian Wood – thank you for those weekly escapes into alternative worlds (of tabletop games). In our group, I found friendship, understanding, passion for adventure, and, above all, acceptance. Working on a PhD is often an isolated and lonely existence. My friends and colleagues made my time at UBC enjoyable, intellectually stimulating, meaningful and fulfilling. Spencer McKay, Dominik Stecula, Charlie Roger and others helped me find my way when I felt intellectually and otherwise stuck. I am especially grateful to Salta Zhumatova and Alex Held for their friendship, and for listening to me during the times of successes and struggles. Deborah Barros Leal Farias and Anastasia Shesterinina have been my friends, confidants and conference hotel roommates. I am grateful for their mentorship, and continue to be inspired by their energy, determination, achievements and kindness. I am, of course, indebted to my family and close friends for their relentless support and tolerance of my irregular work patterns, absent-mindedness and frequently scattered thoughts on issues sometimes too abstract for anyone to care. My gratitude goes to Freddie Hardman Lea for his thoughts, encouragement and unfaltering faith in the strength of my mind. Two of my long-term friends and colleagues have also been crucial to my wellbeing and intellectual growth. Don Munton has always supported my academic dreams and embraced my ideas. Chad Briggs had

xi both warned me about pursuing a PhD and firmly believed I was more than capable to embark on (and complete) this journey. His own work has inspired me to open my mind to new possibilities and innovative ways of thinking. Two more people I met and befriended during my doctoral studies deserve my wholehearted gratitude: Eric Merkley and Dr. Jennifer Gagnon. I don’t think the completion of my dissertation would have been timely without Eric. He taught me how to conduct content analysis and experiments and kept me company during days of mind-numbing coding. Through challenging my ideas and theories, he held my work to the highest academic standard. Without Jennifer, I wonder whether I would have completed my PhD at all. Every time I stumbled, she was there to tell me to keep going. I was always welcome in her home. Her wisdom, determination, confidence and unending faith in me kept me grounded, focused and hopeful. She continues to inspire me in both career and life. Jennifer has told me once that to finish a PhD I only needed to jump through hoops – they did not need to be flaming hoops. Thank you, Jen, for patiently waiting by my side with buckets of water in case I caught fire. A few times I indeed did. I also wish to acknowledge the financial support that I received during my doctoral studies. My work has been funded by SSHRC Vanier Canada Graduate Scholarship, Izaak Walton Killam Memorial Doctoral Fellowship, and the Olav A. Slaymaker Fellowship in Environment. I have also received support from the Department of Political Science and the Faculty of Arts at UBC, and from the Senior Women Academic Administrators of Canada (SWAAC).

xii

To Jennifer.

xiii

Chapter 1. Introduction

Environmental disasters are frequently catalysts for social and political change (Dawson 1996; Pelling and Dill 2010; Hannigan 2012). A growing body of literature has identified disasters as political events that open windows of opportunity for political actors and affect the actions of governments and social movements. For example, disasters may exacerbate or lessen violent conflicts, help governments improve their public image, serve as focal points and spring boards for protests, and even bring down regimes (Chen 2009; Pelling and Dill 2010; Birkmann et al. 2010; Prater and Lindell 2000; Kelman 2006; Brancati 2007; Beardsley 2009; Carlin et al. 2014). Yet, the literature on the effects of disasters as catalysts for political change is inconclusive and at times contradictory (Slettebak 2012). The available empirical evidence is mixed – disasters of similar scale and impact seem to encourage collective action in some cases but fail to do so in others (Flores and Smith 2013). The empirical cases of industrial disasters and related protests, for example, are especially puzzling. Unlike natural disasters that are often described as ‘acts of God’, industrial disasters generally involve perpetrators visible to the public (Birkland 1997, p. 2). While some of these events generate mass nationwide (and international) protests with far-reaching consequences, many others go largely unnoticed and protests, if any, remain small and localized (Macdonald 1980; Lendon and Martin 2007; Anderson and Marhadour 2007, p. 98). For example, the 2010 Deepwater Horizon and the 1979 Ixtoc I disasters top the list of the world’s worst oil spills, yet the former led to widespread media coverage and protests across the United States and the United Kingdom, while the latter resulted in almost no public backlash and “faded into history almost without a trace” (Sindermann 2005, p. 5; Jernelov 2010).1 Similarly, Canada’s 2014 Mount Polley mining disaster, one of the largest in Canadian history, led to only a small localized protest even though the responsible corporation suffered no charges or fines (Linnitt 2015). Japan’s 2011 Fukushima Daiichi nuclear disaster generated almost immediate anti-nuclear protests in Germany, but it took much longer for the Japanese people to take to the streets, and the movement was less organized and had far less political

1 The Ixtoc I spill in the Gulf of Mexico was at the time the “most disastrous oil spill the western hemisphere [had] ever witnessed” (Myer 1984, p. 1). The spill was caused by a Mexican company, Pemex, and lasted from June 1979 to March 1980, contaminating the beaches of Mexico's eastern shore as well as the shore of Texas. The Texas Gulf coast suffered large environmental and economic damages (e.g., tourism and fishing industry losses as well as the cost of clean-up) (Macdonald 1980, note 5). The US government was never successful at getting adequate compensation from Mexico (Macdonald 1980; Myer 1984).

1 traction (The Economist 2014). If disasters are political triggering events, as the existing literature suggests, why do they often fail to generate large scale collective action? In fact, why do some highly damaging industrial environmental disasters succeed, and others fail, to catalyze mass protest movements? What conditions account for this variation? This dissertation examines industrial environmental disasters2 and their effects on protest mobilization. It is a collection of distinct analytical chapters integrated to address the complex issue of protest. In the following six chapters, I present a study of environmental and social impacts of oil spills, mine leaks, and nuclear disasters in terms of their damage to ecosystem services and local economies, and how such damage affects public willingness to participate in nonviolent resistance. Given the complexity of the protest mobilization process, to best answer my research questions I employ various methods and tools, including a geographic information system (GIS) analysis, qualitative comparative analysis (QCA), content analysis, and a survey experiment. Taken together, the dissertation is a comprehensive examination of different factors linked to post-disaster protest mobilization across advanced industrial democracies. I apply both established and new theoretical concepts to better understand empirical patterns in protest movements, viewing them as a form of political action. This chapter serves as a brief introduction to this research; it provides the rationale behind the project as well as the core findings, contribution to knowledge, and outline of chapters. The literature on social and protest movements is vast, with sociologists and political scientists continually improving our understanding of the origins, objectives, workings, and impacts of movements that push for political or policy changes within and across established political systems. While traditionally, social movements were viewed as extra-institutional politics and often condemned as illegitimate, they have played a crucial role in the formation of institutions, regulatory reforms, and even shifts in the prevailing cultures (West 2013, pp. 154– 156). Protest movements are significant, because they signal citizen discontent with the existing socio-political system. Large protest movements threaten government legitimacy and even survival; they attract a lot of attention, which adds to their potential for achieving change. Nonviolent protest movements have been found to be especially politically influential in Western democracies, because they are less costly to participants and therefore often attract wide support (MacAdam, Tarrow and Tilly 2001; Chenoweth and Stephan 2011).

2 The terms ‘industrial disasters’ and ‘industrial environmental disasters’ are used interchangeably. Different types of environmental disasters are discussed in Chapter 2.

Environmental movements in particular are responsible for some of the major transformations in Western societies, from changes in environmental policy and regulations to institutional changes such as establishment of environmental protection agencies (see Wapner 1996; West 2013; Dauvergne 2017; Dauvergne and Alger 2017; Dauvergne and Neville 2011). For example, the Love Canal disaster in the United States redefined the meaning and gravity of the issue of hazardous waste, and eventually led to new legislation, which assigns responsibility for hazardous sites to industry (Hoffman and Jennings 2010; see also Reed 2002). Recently, environmental protests in both Europe and North America have signaled growing public discontent with environmental degradation and related natural resource projects, including fracking, mining and pipeline constructions (Ilie and Sigheti 2013; Halliday 2017; Visser 2017; CBC 2018; see also Neville and Weinthal 2016). Critical events such as major industrial environmental disasters help focus such discontent. While this research centers on environmental disasters, and therefore environmental damage, the protest movements that occur after disasters may not necessarily be driven by environmental activists. As explained in Chapter 2, due to the varying nature of risks and damage that industrial disasters pose, they have a potential to mobilize people across different backgrounds and interests. This research has been built on the work of many social movement scholars, and it utilizes the established social mobilization concepts of grievances, resources, political opportunity structures, and framing. I discuss this overarching theoretical framework and provide a detailed literature review in Chapter 2 (as well as in the subsequent chapters, as applicable). This research is not meant to challenge the relevance of these prevailing concepts; rather, it expands the framework to account for the mixed empirical evidence when it comes to the emergence and growth of protest movements, and particularly those that occur after industrial environmental disasters. Disasters are frequently labelled catalysts (Pelling and Dill 2010; Kelman 2006); their effects are conditional upon some pre-existing social conditions such as economic inequality, regime repression or ongoing violent conflict (Olson and Gawronski 2003; Kelman 2006). Disasters as catalysts can be understood in two ways: 1) critical junctures, and 2) disasters as producing “accelerated status quo” (Pelling and Dill 2010, p. 22). The former treats disasters as historical turning points that create irreversible changes in affected social systems (Olson and Gawronski 2003; Birkmann 2010). For example, according to Hoffman and Jennings (2010, p. 19), such events “can highlight breakdowns or failures of existing institutional arrangements,

3 thereby creating chaotic shifts in the trajectory of institutional development.” The latter view assumes that change is path dependent; therefore, disasters do not change the pre-disaster trajectories but merely speed them up. In this perspective, disasters are viewed as triggers; they are “the actions [or occurrences] that provide the spark that ignites the fuel provided by the underlying [i.e., structural] causes” (Barrington 2012, p. 334). Triggers have been of less interest to social scientists, because they are believed to be substitutable – while some trigger may be necessary for a causal chain to unfold, specific triggers are usually viewed as unimportant. Similarly, disasters are often interesting to scholars as events that produce socio-political or policy changes but not necessarily as events in themselves (Birkmann et al. 2010). This is an indication of a shift from the hazards-disaster tradition in disaster studies where the focus was on attempts to understand the events (hazards) and their characteristics (Burton and Kates 1964; Burton, Kates, and White 1978; Rodriguez, Quarantelli, and Dynes 2007, p. 9). Recent studies of disasters have shifted attention to social change and vulnerability as a social construct (Alexander 2005; Cutter 2005; Pelling and Dill 2006; Rodriguez, Quarantelli, and Dynes 2007, p. 10; Jones and Murphy 2009). Although this research treats disasters as social phenomena, embedded in social relations, it, in part, represents a return to studying physical damage and human behavior in the immediate disaster aftermath. Such a shift allows for a more comprehensive assessment of disasters and their social effects. Even though I do not examine change (only potential for change embodied by protest movements), I adopt the perspective of disasters as critical junctures. Viewing disasters as triggers that merely accelerate status quo implies that their unique characteristics are negligible. To the contrary, some of my findings suggest that such characteristics – and specifically the type of damage and the uncertainty of disaster impacts – themselves shape the post-disaster social dynamic. The analyses and findings presented in the subsequent chapters confirm that grievances, resources, political opportunities, and framing matter in protest emergence and growth. I offer some specific explanations about how and why they matter in the context of environmental disasters, paying special attention to the much-debated concept of grievances (see Pinard 2011). I accept the arguments of scholars who emphasize that we cannot treat the grievance element as an unchanging background condition (Walsh 1981; Kerbo 1982). I identify three types of grievances linked to environmental disasters: those that are suddenly imposed (and therefore represent the element of change), and those that are rooted in the long-standing issues of social

4 justice and environmental justice (and therefore are less likely to change abruptly). The analyses in Chapters 3 and 4 reveal that sudden grievances are likely to be more relevant as a motivating factor than grievances that stem from social and environmental justice issues. This suggests that more theoretical fine-tuning of the concept of grievances in protest mobilization may be able to answer the question of whether they are relevant – or explain why there is such mixed empirical evidence when it comes to their effects. My research also adds the element of uncertainty to the protest mobilization process. Few social movement scholars have fully explored the role of uncertainty in protest despite the emotional (and therefore potentially mobilizing or demobilizing) effects such uncertainty is believed to have on individuals. Uncertainty often generates fear or anxiety, prompting people to avoid risk-taking behaviour such as protest. However, as discussed earlier, major – and often very uncertain – disaster events have had the opposite effect, suggesting that different mechanisms may be at play. Seeking to understand this contradiction between the theory and empirical reality, part of my research explores the political effects of post-disaster uncertainty, and specifically the uncertainty of disaster impacts. Overall, while some of the results presented in the following chapters align with the prevailing literature, others are unexpected. For example, I confirm that different factors enable protest emergence and protest growth – while structural conditions such as long-term grievances and political opportunity structures are largely responsible for protest emergence, framing activities by political actors tend to influence protest size. I also find that post-disaster protest emergence is not influenced by any specific geospatial factor, but population density makes protests easier to grow. My work further reveals that although people seem to value the environment, the reason might be nature’s instrumental rather than intrinsic (or inherent) value. In other words, environmental damage from disasters may increase people’s willingness to participate in protest action, but only if such damage affects areas used (or perceived as useful) for human activities. This suggests that whether sudden grievances serve as a prominent mobilizing factor or not depends on their meaning, which is rooted in the values that individuals attach to the affected environment. I also confirm that certain structural conditions must be in place to enable protest action. While I find only one necessary condition for protest emergence (contrary to some theoretical expectations and literature), I discover combinations of sufficient conditions. Most of these

5 configurations are not surprising, given the context of rich industrial democracies; however, a few – such as the interaction of income inequality with other variables – warrant closer examination. Specifically, although high income inequality should be linked to higher likelihood of protest, the combinations of conditions that lack high inequality create an environment more conducive to the emergence of post-disaster protest movements. The remainder of my research explores uncertainty of disaster impacts as communicated in the news media. Therefore, my focus is on framing as a strategic activity of specific political actors attempting to create shared understandings of reality. Effective framing is needed for both protest occurrence and growth. I specify the post-disaster framing dynamic as one that occurs under the conditions of uncertainty, arguing that such conditions shape the protest mobilization process. I find that whether post-disaster uncertainty has a demobilizing effect depends on people’s pre-existing beliefs, and specifically on their ideological predisposition. More politically liberal individuals resist the dampening effects of uncertainty, while more politically conservative individuals embrace it. This analysis of post-disaster framing represents a novel theoretical contribution to the study of both disasters and social movements. Overall, my research makes three types of contribution to knowledge: theoretical, empirical, and methodological. Theoretically, the dissertation expands our understanding of the protest mobilization process, by adding the concept of uncertainty (as well as relatability, discussed in Chapter 7) and theorizing the types and effects it may have on the likelihood of protest. I view protest as an intermediate step with a potential to lead to larger socio-political changes, but such potential may, for various reasons, not be realized. Empirically, this dissertation relies on data on 38 cases of large industrial environmental disasters from 1900 to present. I have collected and analyzed economic, political, social, geospatial and medial data for a wide selection of countries and years. The media coverage includes newspaper articles in English and German languages. Overall, the dissertation is the first comprehensive effort to go beyond small-N studies of industrial environmental disasters in advanced democracies. Due to the complexity of my overarching research question, I use a mixed-method approach to assess disasters’ social effects, relying on four diverse methods: a GIS analysis, qualitative comparative analysis, content analysis, and a survey experiment. This approach allows for an examination of the complex linkages between environmental disasters and protest movements in a novel way.

In the following chapters, I explore some of the most prominent protest mobilization factors, divided into two broad groups: 1) disaster characteristics, and 2) post-disaster dynamics (i.e., political actors interacting under some structural conditions). Chapter 3 addresses the former, while Chapters 4 to 6 examine the latter. In Chapter 2, I present my overarching theoretical framework along with the categorization and definition of basic concepts. I review the existing social science literature on disasters and provide the reasoning for my classification of disasters according to their origin and effect. Furthermore, I distinguish between different types of industrial disasters, and explain my focus on oil spills, mine leaks, and nuclear accidents. In this chapter, I also define my scope conditions, provide a definition of protest, distinguish protest movements from social movements, and introduce the main factors likely to play role in the emergence of social and protest movements: grievances, resources, political opportunity structures, and framing. Chapter 2 contains only a brief overview of these concepts. The subsequent chapters address them separately and include a broader literature review of each. Chapter 3 examines the first concept likely to be relevant in protest mobilization – grievances. Here, I focus on sudden grievances as a prominent characteristic of disasters, and ask: Why might sudden grievances be linked to protest mobilization? I theorize that such grievances are rooted in two elements: objective losses of environmental values and reactionary emotions, both due to damage to the natural environment, human health, and built environment. I operationalize and examine these concepts in geospatial terms (and through a GIS analysis) – as damage in specific areas that people are likely to value such as national parks or densely populated zones. In Chapter 4, I examine the economic and socio-political contexts that are likely to enable or constrain the emergence of post-disaster protest. These conditions include resources available for social movements as well as the political opportunity structures. I include grievances, but unlike sudden grievances explored in Chapter 3, these grievances are economic and environmental, understood as long-term, underlying social processes. I use a range of variables and proxies to capture these concepts. In a search for necessary and sufficient conditions for the emergence of post-disaster protest movements, I utilize a qualitative comparative analysis (QCA). Chapter 5 examines framing and focuses on ‘the language of disasters’. Theoretically, my goal is to understand how different political actors – activists, corporations and governments –

7 interact under the conditions of post-disaster uncertainty through the use of media frames. This chapter is therefore grounded in a comprehensive literature review of framing research. I discuss the above-mentioned framing actors, their interests, and the most likely types of frames (including the type of tone) they use after disasters. The bulk of this chapter consists of a text analysis of major newspaper coverage in three cases of industrial environmental disasters, selected due to the associated varying protest sizes, from small (the 2014 Mount Polley mine leak) to medium (the 2010 Deepwater Horizon oil spill) to large (the 2011 Fukushima nuclear disaster and its effects in Germany). I find that negative tone frames – and specifically uncertainty and relatability – are linked to the largest protests. To assess the effectiveness of these types of frames in protest mobilization I turn to an experiment presented in Chapter 6. The last analytical chapter in this dissertation examines the effects of uncertainty on the public willingness to protest. Relying on the findings from the content analysis in Chapter 5, I examine psychological, emotional, and ideological variables and the ways in which they shape a willingness to protest. The chapter presents a survey experiment of approximately 3,600 adults in the United States. The results challenge the psychological literature on demobilizing effects of uncertainty and highlight the need for more investigation of the interaction effect between uncertainty of disaster impacts and political ideology. In the next chapter, I turn to the concepts and theories underlying this research as a whole.

Chapter 2. Disasters and Protest: Definitions and Theoretical Framework

What are disasters? How do we conceptualize these events in relation to human societies? How do we study them? Do different types of disasters have different social effects? If so, why and in what ways? This chapter begins a comprehensive effort to examine effects of industrial environmental disasters on ecosystems and economies as well as ways in which disaster damage affects public willingness to participate in nonviolent protest in the disaster aftermath. The chapter discusses disasters in two ways: as events in and of themselves (albeit in relation to the social environment in which they occur) and as events with social effects (i.e., they produce social response which may or may not lead to societal changes). In particular, disasters are treated as political events that may open windows of opportunities for political actors to “entrench or destabilize current power-holders, change power-sharing relationships within recognized sectors, or to legitimise or de-legitimise new sectors” (Pelling and Dill 2006). I examine a specific political feature of disasters – their protest mobilization potential. This chapter serves as a theoretical background to this project; it is divided into two main parts. The first contains a discussion of the existing state of knowledge on disasters in social science and presents a typology of disasters used to clearly delineate the events at the centre of this study. Disasters are categorized as: 1) social vs. environmental, and 2) according to their source and speed of onset. Such relatively simple classification allows for a systematic examination of social impacts of specific types of industrial environmental disasters: oil spills, mine leaks and nuclear accidents. The second part of this chapter contains the overarching theoretical framework for this research – as it relates to post-disaster protest movements and industrial environmental disasters. The chapter concludes with a summary and discussion.

1. Disasters in Social Science

In the prevailing literature, disasters are understood as serious disruptions of societies that often bring widespread destruction, and cause human, material, economic, or environmental losses (Gephart 1984; UNISDR 2004, p. 17; Quarantelli 2005). Broadly speaking, studies of disaster aftermath tend to focus either on physical damage or social effects of disasters. The former describe or assess environmental impacts, property damage and human life loss (Jernelov and Linden 1981; Lagadec 1982; Moldan et al. 1985; Graf 1990; Jernelov 2010; Norse and

Amos 2010; Kerr et al. 2010; White et al. 2012). The latter focus either on popular responses such as activist campaigns, protests and volunteer mobilization (Shaw 2004; Grant 2014) or policy responses, including legal changes and international treaties (Macdonald 1980; Myer 1984; Birkmann et al. 2010; Feldhoff 2014) or both (Molotch 1970; Perez 2003; Hernan 2010; Elliott 2013). The social effects of disasters are, of course, linked to physical damage, but they are also shaped by some pre-existing socio-cultural conditions and social dynamics (Pelling and Dill 2010). This research is predominantly concerned with social – and particularly political – effects of disasters. Disasters were long treated as non-political events and thus largely neglected in political science (Olson 2008; Hannigan 2012, p. 8). They are, however, political in at least two aspects. First, disasters are political events in and of themselves, because the need of the government to not only manage but also explain a disaster to the public opens space for politicization of the event (Pelling and Dill 2006; Olson 2008). Second, disasters produce indirect or secondary political effects (Pelling and Dill 2006). Because they create power vacuums and highlight power failures, they open windows of opportunity for various groups to push through their agendas. As such, disasters may lead to changes in social structures, power arrangements and institutions (Molotch 1970; Gephart 1984; Oliver-Smith 1996; Perez 2003, Hoffman and Jennings 2010). The literature on the politics of disasters has grown over the past decades, albeit slowly. Disasters have been studied in light of their electoral impacts and leader survival (Prater and Lindell 2000; Flores and Smith 2013), effects on pre-existing grievances and legitimacy of regimes (Le Billon and Waizenegger 2007; Chen 2009; Olson and Gawronksi 2010; Pelling and Dill 2010; Carlin et al. 2010), level of regime repression (Wood and Wright 2015), social networks and norms of interaction during emergencies (Gillespie and Perry 1974; Schneider 1992; Aldrich 2014; Metaxa-Kakavouli, Maas, and Aldrich 2018), disaster-related governance (Comfort et al. 2001; Comfort 2002; Robinson et al. 2013), and the politics of disaster aid (Stromberg 2007; Drury and Olson 1998; Cohen and Werker 2008; Wilder 2010). Among the most widely studied topics is the relationship between natural disasters and violent conflict. In particular, scholars are divided over whether natural disasters mitigate conflict by fostering cooperation among states or various groups (Quarantelli and Dynes 1976; Kelman and Koukis 2000; Evin 2004; Kelman 2006; Enia 2008; Kreutz 2012) or exacerbate it because of their negative effects on scarcity and economic development (Ember and Ember 1992; Miguel, Satyanath, and Sergenti 2004; Brancati 2007; Nel and Righarts 2008).

At the domestic level, natural disasters are viewed as focal points and triggers for internal conflict such as protests and revolutions (Albala-Bertrand 1993; Drury and Olson 1998; Brancati 2007; Nel and Righarts 2008; Bearsley and McQuinn 2009; Slettebak 2012). Aside from the conflict literature, intrastate studies of disasters focus on the dynamics/politics of media coverage (Molotch and Lester 1975; Anderson and Marhadour 2007; Chattopadhyay 2012) and disaster responses of various types of political actors such as governments, corporations and activists (Luft 2009; Cherry and Sneirson 2011; Breeze 2012). Compared to theoretical work on the political effects of natural disasters, the existing literature on industrial disasters is less developed. It consists mostly of single-country case studies, largely unconnected in any systematic theoretical way (Molotch 1970; Gephart 1984; Birkland 1998; Fortun 2001; Hasegawa 2014; Niggemeier 2015). Aside from scientific analyses and descriptions of unfolding events (Jernelov and Linden 1981; Lagadec 1982; Moldan et al. 1985; Jernelov 2010; Norse and Amos 2010), industrial disasters are studied in terms of their political or social effects (i.e., disasters leading to changes in socio-political structures and systems) and their policy effects (i.e., disasters as focusing events). Given the breadth of societal changes that disasters may trigger, studies in the former group rarely focus on the same outcome, which contributes to the fragmentation of this literature. Studies of disasters as focusing events are more unified. The primary concern is to understand whether and how disasters change domestic policy agendas (Birkland 1997 and 1998; Busenberg 2001, Bishop 2014). Focusing events are defined as “sudden, attention-grabbing events that help politically disadvantaged groups to push through messages suppressed by dominant groups” (Birkland 1998, p. 53). These events are relatively uncommon, harmful, and known to policy makers and the public more or less simultaneously (Birkland 1998). The theories developed in studies of focusing events, however, fail to explain why some major events become ‘focusing’ while others do not. Furthermore, although they provide a useful general theoretical framework for studying industrial disasters, the studies often neglect the role of these events in protest mobilization. Lastly, focusing events include environmental disasters but also many other types of contingencies with different impacts. This combining of different types of ‘disastrous’ events into a single category is problematic for different reasons, outlined below.

1.1 Disaster typologies

Lack of clear event classification has been a persistent problem in disaster studies. There is a longstanding practice in disaster research to combine different types of events into the same category and label them as disasters: airline crashes, fires, tornadoes, floods, earthquakes, wars and mass kidnappings (Berren, Beigel and Ghertner 1980; Martin 2006). Even when scholars focus specifically on natural or industrial3 disasters, they tend to combine them regardless of their source (Birkland 1997) or speed of onset (Mitchell 1996; Sindermann 2005) or any other clear delineators (Hoffman and Smith 2001; Hernan 2010). Practically, this is problematic, because the lack of a single classification system has resulted in empirical studies of disasters with anomalous findings (see Quarantelli 1987). Conceptually, it implies that different types of disasters have the same social effects or that disaster type has no significant impact on the studied outcome. There is a lack of empirical studies that support or contradict such an implication. Rather than creating a new typology, I rely on a common understanding of basic disaster characteristics. Gephart (1984) recognizes four types of disasters: 1) naturally caused environmental, 2) naturally caused social, 3) socially caused environmental, and 4) socially caused social.4 This project is about environmental disasters – events that damage ecosystems or ecological complexes valued by humans. This is not to say that an airplane crash or a war do not, to an extent, cause damage to the environment, but human life loss is usually the most significant effect of such socially caused social disasters. This research examines disasters that do not directly and immediately result in a large-scale loss of life but have significant environmental impacts. Furthermore, disasters, whether social or environmental, can be classified based on their source, speed of onset, scope of impact, potential for occurrence/reoccurrence, and control over future impact (which is linked to social preparedness) (Berren, Beigel and Ghertner 1980; Barton 1963; Barton 1969; Quarantelli 2000). Focusing on too many characteristics results in overlapping disaster typologies, which, as discussed earlier, has been problematic in disaster

3 These are also sometimes referred to as technological disasters. 4 These are ideal types useful for formulating theoretical propositions. In reality, a disaster may have more than one type of causes and more than one type of effects. 12 studies. Therefore, in this research, environmental disasters are broadly categorized according to their source5 (natural or industrial) and speed of onset (acute or chronic)6 (see Table 1). Natural disasters often arise from natural forces; they result from “unfavourable changes in the environment” (Kondratyev et al. 2002, p. 20), while industrial disasters are linked to some social organization involved in potentially harmful exploitation of natural resources (Gephart 1984; Birkland 1998; Sindermann 2005).7 They tend to be ‘acts of corporations’ rather than ‘acts of God’. Acute (also known as sudden-onset) disasters cause sudden harm (e.g., loss of life, illness) shortly after a single exposure, while chronic (or slow-onset) disasters take a longer time to manifest and are often viewed as social problems involving complex agents (Hannigan 2012, p. 13). The extent of their harm is less clear (Birkland 1998; Adeola 2011).

Table 1. Typology of environmental disasters according to their source and onset. Acute Chronic Natural e.g., hurricanes, typhoons, e.g., desertification, droughts, extreme earthquakes, volcanic eruptions winter conditions Industrial e.g., marine blowouts, oil tanker e.g., leaking pipelines, hazardous waste accidents, nuclear accidents contamination, long-term mine leaks, oil seepage, nuclear testing sites

The focus of this research is on industrial environmental disasters with acute effects8 in OECD countries. Such countries are comparable in terms of regime type, development, and wealth (i.e., they are democracies with high-income economies and high Human Development Index), which allows for controlling for these factors. Acute industrial disasters are theorized to lead to “uncertainty, anger, anxiety, frustration, blame, depression, isolation, a loss of control, and a distrust of governmental authority” (Adeola 2011, p. 19). People then tend to express such

5 In the prevailing literature, disaster ‘source’ is often conflated with disaster ‘type’, which leads to a distinction of disasters as natural vs. humanmade events. Such distinction, however, is too broad, because it includes events such as airline crashes and wars (i.e., these are humanmade). Here the source refers to the source of environmental damage – ‘natural’ destruction of the environment vs. industrial pollution. 6 Rather than dismissing the rest of the common disaster characteristics, this research treats the scope of impact, potential for (re)occurrence, and control over future impact as independent variables. These will be discussed in later chapters. 7 Natural disasters, and especially those linked to climate change, can be anthropogenic in origin. Others, such as earthquakes or volcanic eruptions, are more clearly products of natural forces. 8 This research does not include chronic industrial disasters, because they likely involve different social mechanisms than acute industrial disasters. Contrary to chronic disasters that may be hidden or covered up for a long time by the actors responsible, all actors (e.g., the responsible actor, the government, or the affected public) learn of large acute disasters almost simultaneously (Birkland 1998). This lowers any potential advantage that the corporation or the government has when framing the event. The power structures and dynamics are therefore likely to differ in the aftermath of acute vs. chronic industrial disasters.

13 emotions through collective action – either violent (e.g., riots) or non-violent (e.g., peaceful protest) (Suliman 2014). For reasons outlined below, the variation in the occurrence and size of non-violent protest movements in the aftermath of these events is more empirically puzzling than after other types of disasters. Chronic environmental disasters – whether natural or industrial – have received less attention from social scientists. An exception is the vast literature on the links between climate change, resource depletion and violent conflict (Homer-Dixon 1991; Kaplan 1994; Homer-Dixon 1994; Matthew 2002). However, these studies are in wide disagreement over concepts, data and methodologies, and many of their findings are highly inconclusive (Salehyan 2008; Slettebak 2012; Selby 2014). Studies of chronic industrial disasters are predominantly descriptive accounts of events (Hernan 2010; Adeola 2011). They often result from systemic forces (e.g., chronic environmental stress), and are more difficult to comprehend and react to by the public (Hernan 2010; Adeola 2011). Therefore, varied public response to these disasters is to be expected. Acute industrial disasters that do not provoke public response are much more empirically puzzling. Economic factors such as level of national wealth, development, resource scarcity and income (in)equality have been established as reasonably good predictors of the occurrence and trajectories of civil unrest in the aftermath of natural disasters (Drury and Olson 1998; Brancati 2007; Nel and Righarts 2008; Bearsley and McQuinn 2009). Specifically, high level of national wealth and development, low resource scarcity and low income inequality decrease the likelihood of civil unrest after natural disasters. These factors, however, may be less fitting for cases of industrial disasters in advanced industrial democracies. Empirically, there is a variation in protest movements in the disaster aftermath within rich democratic countries, which may reduce the explanatory power of most of these economic factors.9 Since industrial environmental disasters are conceptually different from natural disasters (i.e., industrial disasters usually have a clear perpetrator to blame, and post-disaster protests are often tied to established anti-industry movements), economic variables may not be the primary and most significant driving forces behind post-disaster protests in advanced industrial democracies. Furthermore, even acute industrial disasters are a heterogenous group; they are likely to have different effects.

9 Income inequality may be an exception.

1.2 Industrial environmental disasters

There are three types of industrial environmental disasters based on the kind of pollution they produce: oil spills, mine leaks, and nuclear disasters.10 Although these events share several characteristics (e.g., they are consequences of complex technical processes and often occur due to some features of the social organization of the extraction and exploitation of resources), they are associated with distinct pre-existing conditions and particularly narratives created by anti- industry and pro-industry groups. Consequently, different types of industrial disasters are likely to evoke different images in the eyes of the public, thus shaping public reactions in the disaster aftermath. Perhaps the most fear-inducing and resented of them all are nuclear disasters. To many, nuclear power generation is a terrifying concept; it is “Hiroshima in one’s backyard” (Bouvier 2016). The yellow and black symbol evokes danger – images of mushroom clouds, deformed human and animal forms, and abandoned buildings with radioactive overgrowth. Anti-nuclear movements are well-established in many Western countries, shaping the intense polarization and political struggle of the nuclear debate (Del Sesto 1980). Nuclear accidents and incidents often lead to spikes of public opposition to nuclear energy (Aldrich 2008; Elliott 2013). One reason for this opposition and fear may be the uncertainty that surrounds nuclear power in general, and many nuclear accidents in particular. The science behind nuclear reactions is an enigma to most, and the secrecy and uncertainty surrounding nuclear disasters not only make data collection and analysis difficult but contribute to loss of public trust in nuclear technology (Elliott 2013; Rose and Sweeting 2016). The invisibility of radiation – the ambiguous imagery, harms expressed as probabilities, and unknown scale and duration of health impacts – contributes to uncertainty and feeds the smoldering public fear (Birkland 1998; Paine 2002; Bouvier 2016). In contrast, oil spills and mine leaks are much easier to understand. Mines seem less threatening, since the impacts of mining disasters tend to be localized and often in remote locations. Mining processes are more straightforward and less complex than nuclear technology. They have “built-in slack in terms of responding to problems”, which gives them less

10 Although the universe of cases for acute environmental disasters also contains chemical spills, these are not included here and will not be examined in this project. The primary focus of this research is on large-scale disasters. Large, environmentally damaging chemical spills are uncommon, and most of the spills that have occurred in the past decades are small compared to large oil spills. Data on chemical spills are available from NOAA’s Incident News database (https://incidentnews.noaa.gov) and CEDRE’s database on spills (http://wwz.cedre.fr/en/Our- resources/Spills/(letter)/default). 15 catastrophic potential (Perrow 1984). Consequently, there is no public fear of mining environmental disasters at the scale of nuclear anxiety. In addition, while prominent in Latin America and parts of Europe, the anti-mining movements are less active in OECD countries (Çoban 2004; Dougherty 2011; Urkidi 2011; Vesalon and Creţan 2013). Similarly, oil – in this case transport of oil whether through pipelines or tankers – is a relatively straightforward process. However, the nature of oil transportation makes oil spill locations less predictable than in the case of mine and nuclear disasters. Although pipelines are stationary, they often stretch for thousands of kilometers, often in remote areas where leaks may go undiscovered for a long time,11 causing considerable damage (see, for example, Noah 1994; Jernelov 2010). Oil tankers, on the other hand, may pose risks to either remote or populated areas or both, depending on their routes as well as unpredictable factors such as weather and human error. Main concerns about pipelines and oil tankers are about the conditions of their safe operation to prevent spills – the public concern depends on the pipeline’s or vessel’s route. For example, much of the public opposition to several major pipeline projects in British Columbia stems from public fears of potential contamination of pristine areas (Hoberg 2013). The issue of oil transportation has also become linked with climate change mitigation. Climate activists concerned with greenhouse gas emissions and advocating for a much more significant shift to renewable resource use have become vocal anti-industry opponents (Hoberg 2013; Grant 2014, p. 14). Nuclear disasters, oil spills and mine leaks may inflict large-scale damage, but it is likely not the damage by itself that affects people’s perceptions and resolve to participate in protest. The prevailing literature on industrial disasters points to several factors that also seem relevant: actors’ ability to create a perception of the event in line with their self-interests (Molotch and Lester 1975; Birkland 1998; Sindermann 2005; Fortun 2004; Anderson and Marhadour 2007; Cherry and Sneirson 2011; Breeze 2012) and socio-cultural context within which the disaster occurs (Molotch 1970; Anderson and Marhadour 2007; Hoffman and Jennings 2010; Adeola 2011). Neither degree of harm nor the location can be separated from the social environment in which a disaster occurs. The concepts of harm, risk and safety are socially constructed; their interpretations depend on the identities and lived experiences of the affected people as well as on the framing efforts of political entrepreneurs (Gephart 1984; Hoffman and Smith 2002, p. 11; Oliver-Smith 2002, p. 25). Therefore, these two factors are unlikely to be significant on their

11 This would, however, be an example of a chronic not acute disaster.

16 own; rather, their salience in protest movements will depend on the ways in which they are framed. This reasoning is in line with the prevailing theories of social movements, which serve as a foundation for this project’s theoretical framework.

2. Social Movements and Post-Disaster Protests

This research is concerned with a specific political effect of disasters – the occurrence or absence of non-violent protest, regardless of whether the protest leads to socio-political changes or not. In broad terms, protest is “joint (i.e. collective) action of individuals aimed at achieving their goal or goals by influencing decisions of a target” (Opp 2009, p. 38).12 Protest, as used in this research, is a type of “direct confrontation with a target” (Opp 2009, p. 39). Protest movements are usually limited in geographic scope (they tend to be local or national) and are rarely sustained for long periods of time, which distinguishes them from social movements (Tarrow 2001).13 However, mass protest movements14 have occurred in the past (Chenoweth and Stephan 2011). Unlike collective behavior that is “shortlived, spontaneous, incidental, or not aimed at mobilizing further support”, mass protest movements are indicative of civil resistance – they are more “purposive, coordinated and sustained in nature” (Chenoweth and Cunningham 2013, p. 273). They also tend to attract wide media attention and are likely to achieve policy change, which makes them politically influential (Amenta et al. 2010). Protest activities can be violent, such as riots and revolts, or non-violent, which include signing petitions, joining in boycotts, and attending peaceful demonstrations (Welzel and Deutsch 2011). Nonviolent protest movements are a significant part of modern politics in advanced democracies where popular pressure is an important source of influence on governments (Norris 2002; MacAdam, Tarrow and Tilly 2001). In democracies, protest tends to be a low-cost activity with low moral, physical, informational and commitment barriers (Opp and Kittel 2010; Chenoweth and Stephan 2011, p. 10). Therefore, we should expect to see frequent protests after large disasters. This is, however, not the case. The protests that arise after

12 Protests can also be viewed as ‘repertoires’ used by established social movements (Dalton, van Sickle and Weldon 2009). 13 These are defined by Sidney Tarrow (2001, p. 11) as “socially mobilized groups engaged in sustained contentious interaction with powerholders in which at least one actor is either a target or a participant.” Social movements also generally strive for fundamental changes in the social order. 14 Mass protests can be understood in terms of a large number of participants or breadth of a movement (i.e., the movement goes beyond local level to gain a national following and participants not immediately affected by the disaster) or both.

17 acute industrial disasters are of both types – some are shortlived and driven by a small group of people protesting a government policy,15 while others involve thousands of people, span months and often affect change. Many large industrial disasters have gone without notice or have been quickly forgotten (see Tables 2 to 4). The aim of this research, as discussed in Chapter 1, is to explain this empirical variation.

Table 2. Large oil spills in advanced industrial democracies, 1900 – present.16

Name Location Date Size17 Pre-existing Protest19 (tonnes, movement?18 thousand) Size (ppl) Scope

Kalamazoo USA, MI July 10 3 YES N/A20 River

Deepwater USA, Gulf of Apr – 645 YES Up to 849,000 Transnational Horizon Mexico July 10 (USA)

Small (UK)

Prestige Spain, Galicia Nov 02 77 YES Up to 200,000 National

Sea Empress UK, Wales Feb 96 72 YES Up to 100,000 National

MV Braer UK, Scotland Jan 93 85 YES N/A

Aegean Sea Spain, A Coruña Dec 92 74 YES 10,000 Local

15 These protests are usually aimed at the responsible corporation/industry or the government (because of its unsatisfactory policies prior to disasters or its poor response to the disaster) or both. 16 The list of oil spills is based on data from NOAA’s incident database (available at https://incidentnews.noaa.gov/raw/index). This database (1957 to present) contains information on oil spills and other incidents for which NOAA has provided assistance (scientific support for response). For a more complete dataset, NOAA’s data were cross-referenced with ITOPF’s oil spill statistics (http://www.itopf.com/knowledge- resources/data-statistics/statistics) and CEDRE’s comprehensive database (1917 to 2015) of oil and chemical spills in waters around the globe (available at http://wwz.cedre.fr/en/Our-resources/Spills/(year)/default). Only disasters that occurred in the OECD countries (members at the time of the spill) were included. 17 These disasters were selected due to their large size. Because it is so well known, the Exxon Valdez disaster serves as the cut off point for spill size (although, compared to other major oil spills, the Exxon Valdez disaster was smaller in size). The Kalamazoo River spill is small compared to oil tanker disasters, but it is considered the largest inland oil spill in US history (Brooks 2014). With respect to nuclear disasters, only those ranked 4 or higher on the International Nuclear Events Scale (INES) were selected. The International Atomic Energy Agency (IAEA) considers anything below level 4 an incident rather than an accident/disaster. 18 See McCormick 1991; Kamieniecki 1993; Diani 1995; Nave 2000; Badruddin 2003, p.110; Mehta 2004, p.38; Jimenez 2007; Buckingham 2008, p.41; Leonard 2008; Swedish Environmental Protection Agency; Curran 2015; Sahin 2015. 19 The data on post-disaster protests were gathered from a systematic search of LexisNexis Academic (notes available upon request). The data on Fukushima in Japan came from Hasegawa 2014. The protest data include several types of nonviolent protest: demonstration, petition, boycott, and activist stunts. 20 N/A here stands for ‘data not available’ due to lack of available information on protests that could be linked to these disasters. If any such protests occurred, it is likely they were very small in size – small enough not to warrant much media attention.

Name Location Date Size Pre-existing Protest (tonnes, movement? thousand) Size (ppl) Scope

MT Haven Italy Apr 91 144 YES N/A

Exxon Valdez USA, AK Mar 89 34 YES Up to 10,000 National

Odyssey Canada, NS Nov 88 132 YES N/A

Irenes Serenade Greece Feb 80 100 YES N/A

Independenta Turkey Nov 79 94 NO N/A

Atlantic Trinidad and July 79 287 N/A N/A Empress/ Tobago Aegean Captain

Betelgeuse Ireland Jan 79 40 YES N/A

Andros Patria Spain, A Coruña Dec 78 60 NO N/A

Amoco Cadiz France, Brittany Mar 78 223 YES Up to 15,000 Local

Hawaiian Patriot USA, HI Feb 77 95 YES N/A

Urquiola Spain, A Coruña May 76 100 NO N/A

Jakob Maersk Portugal Jan 75 88 YES N/A

Othello Sweden Mar 70 60 – 100 YES N/A

Torrey Canyon UK, England Mar 67 119 YES N/A

Lakeview USA, CA Mar 1,230 NO N/A Gusher 1910 – Sept 1911

Table 3. Large mine leaks in advanced industrial democracies, 1900 – present.21

Name Location Date Size (m³, Pre-existing Protest thousand) movement? Size Scope

Mount Polley Canada, BC Aug 14 4,500 YES Small Local

Talvivaara Finland Nov 12 1,200 YES Up to 20,000 National

Kingston Fossil USA, TN Dec 08 4,200 YES N/A Plant

Martin County USA, KY Oct 00 1,200 YES N/A

Aitik Sweden Sept 00 2,500 YES N/A

Baia Mare Romania and Jan 00 100 YES N/A the region

Los Frailes Spain Apr 98 5,000 YES N/A

Tyrone USA, NM Oct 80 2,000 YES N/A

21 The data on mining disasters were initially based on a US Centers for Disease Control and Prevention (CDC) database (available at https://www.cdc.gov/niosh/mining/statistics/content/allminingdisasters. html), which contains all mining disasters in the USA from 1839 to present. However, the database is limited in its focus; it appears to be concerned with human fatalities, not environmental disasters. The Canadian Disaster Database (available at https://www.publicsafety.gc.ca/cnt/rsrcs/cndn-dsstr-dtbs/index-en.aspx) contains data on natural and technological disasters by province between 1900 and present, but, like the CDC database, its focus is on fatalities. Therefore, the mining data were supplemented and cross-referenced with data from other sources (EPA 1997; Mudder and Botz 2004; European Commission 2010; European Environmental Agency 2010; WISE 2017). Only mining environmental disasters with a spill size larger than 100,000 m³ were included. Although small compared to the rest of these mine leaks, the Baia Mare disaster is included, because at the time it was considered Europe’s worst environmental disaster since Chernobyl (BBC 2000).

Table 4. Large nuclear disasters in advanced industrial democracies, 1900 – present.22

Name Location Date INES Pre-existing Protest scale23 movement? Size Scope

Fukushima Japan Mar 11 7 YES Up to 110,000 Transnational (Germany)

Up to 64,000 (Tokyo)

Tokaimura Japan Sept 99 4 YES 2,170 National

Saint Laurent des France Mar 80 4 YES N/A Eaux

Church Rock USA, NM Jul 79 N/A24 YES N/A

Three Mile USA, PA Mar 79 5 YES Up to 107,500 National Island

Lucens reactor Switzerland Jan 69 5 NO N/A

SL-1 USA, ID Jan 61 4 NO N/A

Windscale fire UK, England Oct 57 5 NO NO25

Chalk River Canada, ON Dec 52 5 NO N/A

2.1 Grievances, resources, and political opportunity structures

The social movements literature identifies several conditions needed for collective action, the most fundamental of which are grievances, resources, the existence of political opportunity, a low degree of political constraint, and political leaders that use cultural symbols and social networks to mobilize the masses (Tilly 1978; Tarrow 1994; Keck and Sikkink 1998; McAdam, Tarrow and Tilly 2001; Eckstein 2001; della Porta and Diani 2006).

22 Since IAEA does not maintain a database of nuclear disaster, the list is based on a number of sources: Sovacool 2008; Sovacool 2010; Sovacool 2011; the Guardian 2011. The Fleurus nuclear accident is not included in this database. Even though the accident was ranked 4 on the INES scale, it occurred not at a nuclear power plant but an irradiation facility. Only one person was injured and there was no spread of radiation (the Guardian 2011). 23 The INES scale, introduced by IAEA in 1990, is a tool used to communicate the “safety significance” of nuclear events. The scale consists of seven levels where 0 to 3 are considered incidents, while 4 to 7 are accidents. With respect to the accident levels, 7 corresponds to a major accident, 6 to a serious accident, 5 to an accident with wide consequences, and 4 to an accident with local consequences. The INES scale is available at: http://www- ns.iaea.org/tech-areas/emergency/ines.asp. 24 There is no INES level linked to this uranium mill spill; although, the radiation was reportedly comparable to (or even higher than) that of the Three Mile Island disaster. Over 378,500 cubic meters of liquid and 1,000 mg of solid radioactive mill waste was released into the Puerco River. See Graf 1990. 25 See Arnold 1995.

Grievances are among the most prominent reasons behind people’s willingness to participate in a protest movement (Van Stekelenburg and Klandermans 2013; Opp and Kittel 2010). A grievance is “the feeling of having been wronged, as distinguished from the actual or supposed circumstance, acts, or events that are believed responsible for that feeling” (Giuliano 2011, p. 13). Scholars have established that grievances are a crucial factor in collective action; although, the exact ways in which they matter is not quite agreed upon (Perrow 1979; Walsh 1981; see also Pinard 2011, 36–51). Various deprivation scholars have argued that discontent and collective grievances that give rise to particular social movements are a product of some specific set of structural conditions as well as the perceived gap between expectations and reality (Gurr 1970; Morrison 1971; Gurney and Tierney 1982). Environmental disasters can be linked to several types of grievances, discussed in depth in Chapter 3. In short, these are related to the issues of social and environmental justice but also include sudden grievances, which potentially affect (and lead to mobilization of) non-deprived groups (i.e., groups without long-standing social or other grievance). Resource mobilization theorists have moved away from grievances as the primary component in social mobilization. These scholars believe that some grievances are always present in a population and as such they cannot explain the variation in the emergence of protest movements (McCarthy and Zald 1973, 1977; Zald and McCarthy 1979; McAdam et al. 1996; Rucht 1996).26 Instead, the resource mobilization approach focuses on the central role of resources and internal organization in social movement mobilization (McCarthy and Zald 1973, 1977; Zald and McCarthy 1979). Aggrieved population provides the resources and support for social mobilization. Movement leaders then use a variety of strategies, such as bargaining or violence, to influence authorities. The choice of activity depends on previous encounters and relations with authorities as well as the movement’s ideology (McCarthy and Zald 1977). Unlike traditional grievance approaches, the resource mobilization theory emphasizes the strategic nature of interactions between social movements and authorities. These tactics involve inherent trade-offs and depend on the degree of inter-organizational competition and cooperation. Social movements also utilize society’s ‘infrastructure’ – the media, levels of national wealth, or pre- existing networks and institutions.

26 This view has been criticized by some who emphasize that we cannot treat the grievance element as an unchanging background condition (Walsh 1981; Kerbo 1982). 22

Some scholars of industrial disasters believe that the structure of domestic civil society (and specifically the existence of established anti-industry movements) prior to a disaster has a significant effect on post-disaster protest mobilization (Walsh 1981; Lendon and Martin 2007; Hasegawa 2014). This is in part due to pre-existing resources such as established organizations with staff and legitimacy to express opinions on the issue at hand. Strong established movements (i.e., those with some degree of institutionalization and sufficient resources) are likely to mobilize sympathizers easier and in larger numbers than smaller movements (McAdam et al. 1996; Tilly 2004; Saunders 2013). However, as seen in Tables 2 to 4, pre-existing movements have not always enabled protests after large disasters. Whether they are successful will likely in part depend on the existing political opportunities and constraints. According to political opportunity theory, changes in political opportunities and constraints lead to opening of the windows of opportunity for actors to engage in contentious politics (Tarrow 1996). Political opportunity, as defined by Sidney Tarrow (2011), is “consistent – but not necessarily formal or permanent – dimensions of the political environment or of change in that environment that provide incentives for collective action by affecting expectations for success or failure” (p. 163). Political constraints (i.e., factors that discourage contention) generally refer to the level of repression governments use to prevent collective action. Scholars, including those who study political effects of disasters, believe that a greater degree of state repression (potential or actual) lowers the likelihood of collective action (Drury and Olson 1998; Acemoglu and Robinson 2006), and that political openness (defined in terms of available channels for institutionalized political action) increases the likelihood of protest (Eckstein 2001). Disasters have been identified as factors that facilitate political openings; they have been labelled ‘threshold events’ or ‘focusing events’ or ‘critical junctures’ that can lead to organizational and institutional changes (Olson and Gawronski 1985; Birkland 1998; Olson 2008; Birkmann et al. 2010; Pelling and Dill 2010). Such ‘critical events’ often attract the attention of social movements that can magnify or exploit them, and therefore shape political opportunity structures (Meyer and Staggenborg 1996). Assuming the pre-existing environmental or anti-industry movement mobilizes protesters primarily through framing efforts, the effects of grievances, resources and political opportunity structures on post-disaster protest mobilization cannot be fully understood without scrutinizing actors’ framing strategies.

2.2 Framing and post-disaster protest mobilization

Although disasters may impose sudden grievances and open windows of opportunity for political actors to push through their agendas, protest movements do not automatically arise from these conditions. The additional crucial element is a deliberate construction of specific meanings by political entrepreneurs who attempt to impose their views on others in face of competing claims (Tarrow 1996, p. 110; Gephart 1984; Birkland 1997, p. 10). Protest participation is a result of mobilization process, which can be broken down into consensus and action mobilization (Klandermans 1984). The former is about participation because of shared interpretation of who should act, for what reasons, and in what ways. The latter consists of four steps: sympathizing with the cause (which is the result of consensus mobilization), knowing about the upcoming event, and having both willingness and ability to participate (Van Stekelenburg and Klandermans 2013). Each step divides the population into potential participants and those who drop out (see Figure 1) – sympathizers vs. non- sympathizers, targets of mobilization attempts vs. those not targeted, targeted sympathizers who are motivated to participate vs. those who are not, and motivated targeted sympathizers able to participate vs. those who are not. Since potential participants drop out at each step, large protests are likely to develop only when the pool of sympathizers is sufficiently large. The size of this pool is largely determined by consensus among sympathizers, which is achieved through framing (Cooper 2002; Opp 2009, p. 216).

Figure 1. Protest mobilization process.

Source: Van Stekelenburg and Klandermans 2013, p. 896.

In the comparative literature on social movements, framing is understood as “the conscious, strategic efforts by groups of people to fashion shared understandings of the world and of themselves that legitimate and motivate collective action” (McAdam, McCarthy and Zald 1996, p. 6). In simple terms, frames are means of messaging used to construct a type of reality (or one’s understanding of a particular issue) for specific purposes; they can be words or images (Corrigall-Brown 2012; Corrigall-Brown and Wilkes 2012). Three general types of frames are used by activists in social mobilization: diagnostic, prognostic, and motivational. Diagnostic frames are linked to the narratives of injustice; their main purpose is to focus blame or responsibility. Prognostic frames propose a solution to the problem as well as some strategies for ‘a plan of attack’. Motivational frames encompass ‘calls to arms’; they provide rationale for participating in a social movement through construction of “appropriate vocabularies of motive” (Benford and Snow 2000, p. 617). All three types can be identified in post-disaster movements, with diagnostic frames being perhaps the most prevalent (Bucher 1957; Drabeck and Quarantelli 1967; Gephart 1984; Neal 1984; Gephart 1993; Javeline 2003; Waugh 2006; Catino 2008; Pantti and Wahl-Jorgensen 2011). Framing, and particularly the symbolic power of disasters in political discourse, has been a prominent object of disaster studies in social science (Button 2002; Hoffman 2002; Pelling and Dill 2008). Much of the focus has been on the role of news media in framing disasters (Molotch and Lester 1975; Benthall 1995; Perez 2003; Anderson and Marhadour 2007). Through use of symbols that carry emotional weight, media are believed to help activist groups draw attention to certain issues (Birkland 1998). However, many more political actors take part in post-disaster framing: the actors responsible for the disaster, communities affected by the disaster, government regulatory agencies, the scientific community, and various interest groups (Molotch 1970; Birkland 1998; Sindermann 2005). This research takes into account these varying actors and their likely interests as they interact in specific social environments. As a whole, the following chapters are the first comprehensive effort to understand the social effects of industrial environmental disasters through examining all major elements of social mobilization: grievances, structural conditions, and framing.

Conclusion

This chapter provided an overview of the existing state of literature on social effects of disasters as well as an application of the prevailing social movement concepts on post-disaster protest mobilization. While the literature on natural disasters is reasonably well developed (albeit with some unresolved debates), studies of industrial disasters are mostly scattered and fragmented as a whole. Disasters are believed to be catalysts to social or policy changes, with their effects dependent on some pre-existing social, economic and other conditions. Many findings and conclusions about such effects come from studies of natural disasters and violent conflict in developing countries. Similarly, intra-state studies of disasters tend to focus on violent conflict rather than non-violent resistance. These may not be applicable – in whole or in part – on industrial environmental disasters and their effects on non-violent protests in rich industrial democracies. This study makes theoretical contributions to this area of research in general, and to the existing knowledge on industrial environmental disasters in particular. As a first step, this chapter discussed the types of disasters under study, the dependent variables (i.e., occurrence and size of post-disaster protest) as well as the ways in which disasters relate to the fundamental elements of protest mobilization process. In the end, all these elements – grievances, resources, political opportunities, and effective framing – are likely necessary for post-disaster protest movements to develop and grow. This research examines the specific ways in which they matter. Do certain disaster characteristics – type and location of impact, and the nature of uncertainty they produce – have effects on post-disaster protests? Are environmental disasters linked to distinct sets of values, attitudes and grievances? What type of social environment is conductive to post-disaster protests? How do political actors talk about environmental disasters and does their language have any effect on post-disaster willingness to protest? If so, why? To provide some answers, the following chapters systematically examine the discussed social movement concepts as they relate to industrial environmental disasters: grievances linked to the degree and location of disaster damage (Chapter 3), resources and political opportunity structures (Chapter 4), as well as framing and counter-framing of disasters by different political actors (Chapters 5 and 6).

Chapter 3. Disaster Damage and Sudden Grievances: Mapping Environmental Disasters and Protest

“The first law of geography: Everything is related to everything else, but near things are more related than distant things.” Waldo R. Tobler (1970)

Most large-scale industrial environmental disasters are met with very little public response. Empirically, there is a variation in post-disaster protest occurrence and size even if controlled for disaster size, and both within and across disaster types (see Tables 2 to 4 in Chapter 2). This suggests that other characteristics of industrial disasters are likely to be more significant for protest mobilization. This chapter examines one such characteristic – the damage the disaster causes, and specifically its location as it relates to features of the environment that people value. Why does disaster damage matter in post-disaster protest mobilization? Since disasters are social phenomena that affect human societies, the significance of disaster damage can be linked to two grievance-generating factors: 1) damage to the natural environment that people value; and 2) damage to human health, life, and built environment. Factors that potentially constrain public willingness to protest in the aftermath of disasters (e.g., the proportion of population employed by the industry, generational variables such as certain age groups available for protest, and wealth) are not addressed here. These factors are not characteristics of disasters, and they will be discussed as structural conditions in Chapter 4. Several hypotheses about the relationship between the location of damage and protest are proposed in this chapter, and subsequently evaluated through a geospatial analysis. This analysis is an analytical probe to establish whether there are any patterns between disaster location, nonviolent protest, and several geo-spatial variables linked to the values that people hold for the environment. Although the results are mixed, population density in general and oil spills in particular seem to be the two factors most commonly linked to post-disaster protests. I also find that while the loss of environmental value matters in protest mobilization, it is not sufficient for protest to develop or grow. This is in line with the theoretical expectation that grievances alone are not enough to mobilize people for protest. However, some of my findings go contrary to scholars who argue that grievances have little explanatory power in protest mobilization, suggesting that a closer examination of grievances may be warranted.

This chapter is divided into two main parts. The first discusses the concept of sudden grievances, several grievance-generating factors linked to industrial environmental disasters, and hypotheses derived from those factors. The second part consists of a geographic information system (GIS) analysis as a preliminary evaluation of the proposed hypotheses based on the available geo-spatial evidence. It contains a discussion of the significance of using GIS in social science research, means of data collection and visualization, as well as analysis and discussion of the findings. The chapter concludes with a summary and implications of the findings as well as suggestions for future work.

1. Environmental Disasters and Grievances

Grievances can be linked to industrial disasters, because such events frequently have clear perpetrators, and occur in the context of existing (or lacking) government regulations. In other words, if a destructive disaster is preventable, it is likely to lead to public feelings of being wronged – either by the polluter, the government, or both. Three types of grievances can be linked to environmental disasters: issues of social justice, issues of environmental justice, and sudden grievances. Environmental disasters are likely to channel (rather than generate) grievances rooted in issues of social and environmental justice. The former may encourage mobilization by economically and socially marginalized communities protesting a variety of issues, including inequality, poverty, crime or political exclusion (Pelling and Dill 2006). For example, the larger and more widespread the income inequality, the larger the related grievance and therefore potential for protest mobilization. However, in such cases, disasters and their impacts likely serve as a pretext (or a trigger) rather than a true reason for protesting. Similarly, grievances related to environmental justice concern the issue of inequitable share of environmental ills with which poor communities, indigenous communities, and communities of colour live (Kuehn 2000; Schlosberg 2004; Carruthers and Rodriguez 2009; Martinez-Alier 2016). Inequitable share of environmental ills often mirrors the inequity in socio- economic and cultural status. Therefore, a disaster may disproportionally impact certain groups relative to others (Drury and Olson 1998; Adeola 2011), thus exacerbating existing environmental and socio-economic problems. In such cases, a disaster serves as a catalyst to

28 public demands for potentially broader societal changes, but these demands are more directly linked to the state of the environment than to grievances rooted in issues of social justice. There is a third type of grievances that differs from a traditional understanding of the concept. These so-called ‘sudden grievances’ potentially affect (and lead to mobilization of) non- deprived groups; they can be ‘suddenly realized’ (in cases of chronic industrial disasters) or ‘suddenly imposed’ (in cases of acute industrial disasters) (Walsh 1981). Sudden environmental grievances are crosscutting; they can affect a lot of people irrespective of wealth, education or class as long as those people value the affected location. This type of grievance is the main focus of this chapter. Issues of social and environmental justice are often long term, underlying conditions (i.e., structural conditions) rather than specific features of disasters. Why might sudden grievances increase public willingness to protest? The motivational effect of sudden grievances rests on two factors: objective conditions and felt sentiments (see Pinard 2011, p. 5).27 Disaster damage represents loss of environmental values, and therefore creates objective conditions. Furthermore, sudden events often lead to reactionary emotions such as anger, annoyance, or fear. According to Pinard (2011, pp. 91–4), the motivation to participate in protest is rooted in such emotions as well as in external incentives (e.g., altruistic sense of duty or self-interested benefits) and the expectation of success. The next two sub-sections discuss two types of objective conditions and related felt sentiments brought upon by industrial environmental disasters: loss of environmental values from damage to the natural environment, and impacts on human health, life and economies due to disaster proximity.

1.1 Grievances and environmental values

A disaster that damages the environment will likely generate anger and frustration because of affected people’s perceptions that the environment’s value has been diminished. Two types of values are likely to be most affected: 1) market values (related to economic concerns for livelihood) and non-market values (linked to the values inherent in nature that are not traded in the markets). Market values of ecosystem services are ‘prices’ for natural capital as they are traded on commodity markets. If disasters diminish nature’s ‘ecological yields’, the resulting loss

27 Pinard (2011, p. 5) defines objective grievance as “disadvantageous conditions or morally objectionable situations, actual or anticipated, as perceived individually and collectively by members of a social group.” The felt grievance refers to “experienced sentiments of discontent about such actual or anticipated conditions or situations evaluated as unjust or illegitimate and attributable to some responsible agents.”

29 can be expressed in monetary terms. For example, the Deepwater Horizon spill resulted in large revenue losses to the Gulf of Mexico’s commercial fishing industry, estimated at $94.7 million to $1.6 billion in the first eight months (Schleifstein 2016; see also IEM 2010). Non-market values include consumptive and non-consumptive use value (resources and recreational activities that do not enter the market), option value (of having options to choose from among competing alternative uses of a natural environment in the future), existence value (derived from knowing that nature exists), and bequest value (altruistic value of preserving nature for future generations) (Greenley, Walsh, and Young 1981; Haab and McConnell 2002). Environmental disasters diminish any of these values – some are provided as examples in Table 5. Although valuation of natural capital has been criticized for putting a price tag on nature (Monbiot 2011), it is necessary in policy making that involves monetary trade-offs in resource allocation. ‘Pricing’ nature also allows for comparing environmental values over time and across societies. Valuation is therefore useful in understanding and measuring the values that the public holds for the environment.

Table 5. Environmental values and pollution from industrial disasters.

Type of Value Area Species Disaster (example) Market Forest harvested for Commercial fish and Chernobyl – impacts on timber; access to seafood; commercial the reindeer industry in parks; ecotourism livestock Sweden (Soderqvist 2000) Non- Consumptive Resources (e.g., Recreational hunting Deepwater Horizon – market firewood for and fishing recreational fishing personal use) (Alvarez et al. 2014) Non- Recreational Birdwatching Exxon Valdez – nature consumptive activities (hiking, viewing (Hausman, boating) Leonard, and McFadden 1995) Option Water quality for Yet undiscovered Loss of option value due recreational medicinal plants to water contamination purposes; available in the future from mining (Greenley, groundwater Walsh, and Young quality 1981) Existence Protected areas Protected/at risk Prestige – oiled habitats species and fauna (Loureiro, Loomis, and Vazquez 2009) Bequest Protected areas Protected/at risk Prestige – avoid future species oiled species and habitats (Loureiro and Loomis 2013)

As seen in Table 5, several studies have priced the market and non-market costs of various industrial disasters. For example, natural resource damage from Chernobyl in Sweden was estimated at SEK 736 million ($165 million in 2017 dollars) (Soderqvist 2000).28 Losses to recreational fishing after the Deepwater Horizon spill were approximately $585 million (in $656 million in 2017 dollars) (Alvarez et al. 2014). Spanish willingness to pay for oil spill prevention program to avoid environmental losses from future spills like Prestige was EUR 574 million ($876 million in 2017 dollars) (Loureiro, Loomis, and Vazquez 2009). These studies, of course, do not show whether willingness to pay is equated with willingness to protest. However, some studies have linked the diminishment of environmental values to emotions (Kahneman and Knetsch 1992; Leon et al. 2014), which are at the core of social movements (Keck and Sikkink 1999; Wood 2003; see especially Goodwin, Jasper, and Polletta 2001). Emotions, and particularly fear and anger as they relate to risk perception, are also central to understanding grievance-generating effects of disasters that occur in proximity to human populations (Rasmussen 1992; see also Pinard 2011).

1.2 Grievances and disaster proximity

Aside from generating sudden grievances due to loss of environmental values, industrial disasters concern the public when they occur in people’s ‘backyards’. There are several likely reasons why: the ‘NIMBY syndrome’, impacts on human health, human life toll, and property damage and other economic losses.29 Public resistance to industrial development and its negative consequences has been well- examined in studies of the so-called NIMBY syndrome.30 The NIMBY phenomenon involves the public protesting different types of land use – from sites to store hazardous chemical or nuclear waste to low income housing and airports to wind power generation stations – because such developments are considered undesirable (Armour, 1984; Weisberg, 1993; Rabe, 1994; Wolsink,

28 All values were converted (if necessary) using historical exchange rates (available at https://www.ofx.com/en- us/forex-news/historical-exchange-rates/) and inflated with a US Inflation Calculator (available at http://www.usinflationcalculator.com/). 29 These may include losses in non-market values of ecological goods and services, as discussed in section 1.1. 30 The syndrome is defined as “intense, often emotional and usually organized opposition to siting proposals that residents of a local community believe will result in adverse impacts” (Wexler 1996).

1994; Guo 2015).31 NIMBY is usually discussed in the context of unwanted land use, but the underlying contention is also applicable to industrial disasters that occur in close proximity to human populations – the public does not want to be disproportionally affected by negative consequences of technological processes that, in general, benefit a wider population. For example, oil is still needed for most people’s daily lives, but only a fraction of the population usually suffers the consequences of a spill. A variety of reasons for NIMBY syndrome have been proposed – from embracing new environmental values (Thomashaw 1995) to fear of new technological risks (Slovic 1987) to public distrust of authorities (Kraft and Clary 1991). When it comes to industrial disasters, much of the dissatisfaction is likely due to perceptions of high risk due to the toll on human health and life. Nuclear accidents, for example, produce both short and long-term health effects that may reach well beyond the disaster origin. Exposure from inhalation and both ground-level and atmospheric external pathways as well as due to ingestion of contaminated water and food is likely to lead to a spike in illnesses and cancer-related mortalities (Anspaugh, Catlin, and Goldman 1988; WHO 2012). In addition, long- term psychological effects such as fear, anxiety, and depression have been linked to large nuclear disasters such as Chernobyl and Fukushima (Bromet 2012; Hoeve and Jacobson 2012). In cases of oil spills and mine leaks, inhalation and touching of oil products (e.g., during clean-up) and ingestion of contaminated food can lead to adverse health effects (Lyons et al. 1999; Solomon and Janssen 2010). Although less frequently than large natural disasters, industrial environmental disasters also claim lives. For example, the 1972 Buffalo Creek mining disaster resulted in at least 118 fatalities, destroyed 500 homes, and displaced 4,000 people (Davies, Bailey and Kelly 1972, p.1). The event was a massive flood caused by a collapsed coal slurry dam in West Virginia. Other similar disasters have resulted in human casualties: a burst dam at the Ajka alumina plant in Hungary killed ten and injured 120 people (Taylor 2011), and a dam failure at the Tashan iron ore mine in China claimed 254 lives (BBC 2008). Industrial disasters also result in property damage and economic losses not necessarily or solely linked to nature. The ten most damaging industrial accidents between 1900 and 2017

31 The central problem of these developments, as seen by the affected residents, is the distribution of costs and benefits such as effects on human health vs. economic benefits – while the costs may be geographically concentrated, the benefits are usually enjoyed by a wider, more dispersed population. 32 caused economic losses32 between $541 million33 ($1.3 billion in 2017 dollars) and US$20 billion ($22 billion in 2017 dollars) (CRED).34 For example, the Prestige oil spill resulted in US$10 billion ($14 billion in 2017 dollars) worth of damages, and the Chernobyl nuclear disaster caused US$2.8 billion ($6.2 billion in 2017 dollars) in economic losses. These and other grievance-generating factors serve as a basis for development of several hypotheses on disaster characteristics and protest, discussed in the next sub-section.

1.3 Hypotheses

The aim of this chapter is to find out whether there is a relationship between the location of disaster damage and the occurrence of post-disaster protests. The hypotheses are derived from the theoretical insights discussed in the previous two sub-sections; they rely on five independent variables (IVs) and two dependent variables (DVs). The IVs relate to disaster damage in terms of the location of pollution. They are specified as follows:

• Presence/absence of pollution in protected areas such as designated parks or otherwise areas of conservation value; • Presence/absence of pollution that affects designated species at risk or charismatic species; • Presence/absence of pollution of major waterways and/or groundwater; • Presence/absence of pollution in densely populated areas; • Presence/absence of pollution in recreational areas near human dwellings such as public beaches and parks.

Pollution is defined as a volume of contaminants (i.e., chemical substances from industrial processes) that enter the environment and change it adversely. In this research, pollution is understood in two ways: type (i.e., oil, mine, and nuclear pollution) and volume (i.e., the spilled substance measured in tonnes or cubic meters).35 The three types of pollution

32 The economic losses include damage to property, crops and livestock. 33 This was the San Juanico disaster – a series of explosions at a liquid petroleum gas tank farm in San Juanico, Mexico on 19 November 1984. 34 The Deepwater Horizon disaster. 35 An additional way to think about pollution is in terms of its scope (i.e., the size and type of affected area). The scope of pollution differs for each disaster in the database. Therefore, it is not feasible to create consistent

33 examined in this chapter – oil, mine waste, and radioactive material – are point source pollution (i.e., from single identifiable sources). Since all selected disasters are large in volume and linked to varying post-disaster public response, disaster size is likely irrelevant to occurrence and size of post-disaster protests and will not be examined in this chapter.36 A description of the specific independent variables (i.e., pristine areas, species at risk/charismatic species, major waterways/groundwater, population density, and recreational areas) is provided in section 2.2. The dependent variables are protest occurrence and protest size. The former is a binary variable, either present or absent. The latter is an ordinal variable with three categories: small (1– 999 participants), medium (1,000–99,999), and large (100,000–1,000,000).37 Measuring this DV entailed counting the estimated maximum number of individuals that participated in relevant protest events. It is likely that all the IVs are positively correlated with post-disaster protest. Since other factors than grievances are likely needed for protest movements to develop, we may see no protests even if the IVs are present, but we should see no protest if they are absent (e.g., in cases of disasters far away from population centers and away from protected areas). Furthermore, protest size is unlikely to be linked to disaster characteristics. Large protests require a large initial pool of sympathizers who are mobilized through framing. Disaster damage may provide fuel for framing efforts but by itself is unlikely to determine protest size. Therefore, there may or may not be a relationship between protest size and the IVs. Three specific hypotheses follow:

H1: The closer the disaster to dense population centers such as cities, the more likely are protests to occur. Therefore, given their common locations, nuclear accidents are likely to be linked to large protests, mine leaks to small protests, and oil spills to small, medium or large protests, depending on the spill remoteness.

categorization of scope of damage. Instead, this type of impact can be mapped to allow for a more accurate analysis of disaster damage than simply using points of origin for disaster locations. Mapping of the scope of pollution is currently in progress for selected disasters. 36 However, different types of oil and mine waste are likely to cause varying degrees of environmental damage (see, for example, EPA 1995; Dunford and Freeman 2001). For example, different types of oil (e.g., gasoline, crude oil or heavy fuel oil) dissolve or evaporate at varying rates and therefore result in different degrees of harm (McCay et al. 2004). These differences in pollution types are presented in Table 30 in Appendix 1. 37 According to Chenoweth and Stephan (2011, p. 32), an average nonviolent campaign has about 200,000 participants. This is, however, with respect to political resistance and campaigns aimed at overthrowing the established regimes. Environmental campaigns in democracies tend to be smaller, depending on the country (see Rootes 2003).

Human proximity to a disaster is likely to generate public concerns over health impacts, property damage, and/or concerns over losses of some environmental values – for example, aesthetic or recreational. The more densely populated the impact zone, the larger the potential pool of affected (and concerned) individuals, which increases the likelihood of protest.

H2: The greater the loss of environmental values (understood in terms of market and non-market values of the affected environment), the more likely are protests to occur.

Concerns over environmental losses may be linked but not necessarily limited to human proximity to a disaster. For example, individuals have been shown to value charismatic species such as whales and bald eagles even if they live nowhere close to those species’ habitats (Richardson and Loomis 2009). Similarly, protected areas such as national parks may be valuable to some due to an opportunity to visit (if one lives nearby) but also for other reasons – as national heritage, for example, which may be important to individuals who have never visited. Therefore, even damage to remote protected areas may generate sudden grievances and public protests.

H3: There should be no protests if a disaster occurs in remote areas with low environmental value.

If the disaster impact zone is away from dense population centers and protected areas, the damage is unlikely to generate sudden grievances. Therefore, in such cases, there is no immediate reason to expect post-disasters protests to develop. The above hypotheses are rooted in two different understandings of the human relationship with the natural environment. The first is the instrumental view. From this perspective, the environment only has value when perceived in relation to human societies. As such, environmental losses only translate into grievances if those losses directly affect humans and human activities. The second, opposing belief holds that environment has intrinsic (or inherent) value regardless of humans and human needs. Research shows that a lot of people believe in intrinsic value of nature, but it is unclear whether such beliefs are more powerful in motivating actions like protest (Nelson, Bruskotter, and Vucetich 2015). While the analysis presented in this chapter cannot provide a definite answer, an evaluation of the above hypotheses

35 might offer some evidence one way or another. Hypothesis 1, for example, is rooted in the instrumental logic, while hypotheses 2 and 3 rely in part on the recognition of the intrinsic value of nature. The rest of the chapter discusses an evaluation of these hypotheses through a geographic information system (GIS) method, along with an explanation of the significance of this approach, and the process of data collection, coding, and mapping.

2. GIS Analysis

2.1 GIS, social science, and disasters

A geographic information system (GIS) is a computer-based database system and mapping technology used for capturing, storing, analyzing, and displaying geographically referenced data (Batty 2003; Hanewitz 2012). Through GIS one can display multiple layers of different data on a single map, which allows for analyzing patterns and relationships. For example, GIS can calculate distances and area sizes, and layer thematic maps vertically (e.g., layer population density over a city map), or display data for one area over time (e.g., fluctuations in air quality or changing political party affiliations) (Mitchell 1999; Batty 2003; Maantay and Ziegler 2006). GIS is therefore a powerful data collection and analytic tool, with users in both private and public sector – from utility companies, real estate firms, and retail stores (Somers 2004; Fritz and Skerfving 2005; Hackbarth and Mennecke 2005) to health care professionals, law enforcement officials, and various government agencies (Jenks and Malecki 2004; Brown 2005; Craglia, Haining and Wiles 2000). In academic research, the use of GIS and spatial analysis has been prevalent in earth sciences, geography and regional science, but few studies have attempted to integrate spatial analysis into social science (Goodchild et al. 2000). GIS has been used in economic geography to study spatial imprints of trade and development (Arthur 1989; Krugman 1991 and 1996; Ioannides 2000), and in sociology to link individual behavior to a spatial context (Morenoff and Sampson 1997; Sampson, Morenoff, and Earls 2000). In political science, GIS has found application in analyzing voting behavior and patterns (Horn 1999), impacts of policies on specific populations (Sui and Hugill 2002), and geopolitics in general (Starr 1991). Increasingly, GIS has also been used in qualitative research to capture local knowledge and citizens’ concerns

36 and encourage citizen participation in planning and local politics (Dennis 2006; Rambaldi et al. 2006; Wridt 2010). Despite its obvious utility in environmental studies at both local and global levels, the use of GIS has been limited to analyzing relationships between human activities and local environmental changes (e.g., deforestation, land use patterns, or spatial dimensions of sustainability) (Chomitz and Gray 1995; Bockstael 1996; Nelson and Hellerstein 1997; Stonich 1998; Mas et al. 2004; Xiao et al. 2006). The relationship between human activities and global environmental changes has not been fully articulated, with disconnect between earth system science research and human effects such as migration, urbanization, and others (see O’Neil, Weinthal, and Hunnicutt 2017). Similar disconnect is evident in disaster studies, as well. GIS has been used for disaster response and planning purposes as well as to map and analyze impacts of some disasters (Turner 2003; Ivanov and Zatyagalova 2008; Joyce et al. 2009). Yet, to date, no comprehensive analysis of large-scale industrial disasters and their social impacts has been conducted. This study is a step in that direction.

2.2 Data, analysis, and mapping

The database used for this GIS analysis contains 21 oil spills, eight mine leaks, and nine nuclear accidents (see Tables 2 to 4 in Chapter 2). The information on the disaster occurrence and location was collected from several databases, including NOAA’s Incident News, ITOPF’s oil spill statistics, and CEDRE data on oil and chemical spills as well as various governmental, scholarly, and media publications (see Tables 2 to 4 in Chapter 2). The initial GIS data collection and classification was completed in three steps. First, all disasters were located through their coordinates38 through ArcMap, part of the ArcGIS software. The different disaster types were labelled in ArcMap, using simple graphics. Second, protest was coded as an ordinal variable and ranked from zero (none occurred) to three (large protest). The categories, as shown in Table 6, are based on protest population ranges (i.e., counts of protest participants). Overall, the protest data include the locations of protest occurrence, the counts of participants as well as several types of nonviolent protest: demonstration, petition, boycott, and activist stunts.39 The data were collected through a systematic search of LexisNexis Academic

38 Most coordinates are available in NOAA’s Incident News database, but some were obtained from GeoHack and GoogleMaps. Consequently, some of these coordinates may be less accurate. 39 These were counted as separate protest events if different types occurred concurrently.

37 and, in some cases, supplemented by scholarly literature (Elliott 2013; Hasegawa 2014). Protest points were then located on the ArcMap and presented as graduated symbols based on their assigned codes (i.e., size).

Table 6. Coding post-disaster protest data.

Protest population range Label Code 0 none 0 1–999 small 1 1000–99,999 medium 2 100,000–1,000,000 large 3

Third, data from national statistical agencies and other sources were used to create several layers on ArcMap, to correspond with the independent variables identified in the previous section. Specifically, the protest data were overlaid with those on disaster origin as well as the data on population density and protected areas. The latter has been published by the United Nations Environment World Conservation Monitoring Centre and is available as part of the World Database on Protected Areas (WDPA).40 The database contains global data on marine and terrestrial protected areas as defined by the International Union for Conservation of Nature (IUCN) and the Convention on Biological Diversity (CBD).41 These include national parks, wildlife protection areas, ecosystem reserves, conservation areas, green corridors, and protected water surfaces (and the common fishery area in Japan), and are available for download as shapefiles. The data are updated monthly – those used for this analysis were published in November 2017.42 The population density data were more difficult to acquire, since each country collects and publishes this information in their own way. The data for Asia are from Harvard University’s JapanMap and ChinaMap online platforms.43 The European census data are available from the European Commission’s Eurostat database.44 However, since it was difficult to work with this dataset in ArcGIS (due to its large size), an alternative dataset was acquired from ESRI’s ArcGIS

40 Available at https://protectedplanet.net. 41 See http://www.biodiversitya-z.org/content/protected-area. Proposed protected areas are excluded from this dataset. 42 For information on data collection, see https://protectedplanet.net/c/calculating-protected-area-coverage. 43 See http://worldmap.harvard.edu/japanmap/ and http://worldmap.harvard.edu/chinamap/. 44 See http://ec.europa.eu/eurostat.

38 online platform,45 which offers the same data at a lower resolution. For North America, the data sources were Statistics Canada and the US Census Bureau.46 The census population data for Mexico, used in this analysis, have been assembled by an independent researcher and made publicly available online.47 Since the timeline for the disasters as a whole spans a century, displaying the census data for each disaster year in a single map along with other variables was not feasible. Therefore, the maps display the latest available population density data: 2010 for Japan and the United States, 2016 for Europe, and 2017 for Canada and Mexico.48 Due to difficulties with data collection, three independent variables, as specified in the hypotheses, have not yet been mapped: species at risk/charismatic species, important waterways/groundwater, and recreational areas near population centers (e.g., public beaches and city parks). For the remaining variables, a variety of tools, including TextMate, Sublime Text, and Microsoft Excel, were used to consolidate and input the different data formats into the ArcMap. Finished maps were then fine-tuned in Adobe Illustrator. Figures 2 to 4 present the results of these data collection and mapping efforts. To reduce visual clutter, the data were divided regionally and presented as three maps: Europe, North America, and Asia. In each map, the respective disaster symbols mark the geographic origin of the particular disaster. Each is then linked, through solid lines, to its associated protest. As the primary independent variables, both the population density and protected areas layers are also displayed.

45 See https://www.arcgis.com/home/item.html?id=cf3c8303e85748b5bc097cdbb5d39c31. 46 For the US Census data, see https://www.census.gov/geo/maps-data/. The Statistics Canada data are available through Simply Analytics from www.simplyanalytics.com. 47 See https://blog.diegovalle.net/2013/06/shapefiles-of-mexico-agebs-manzanas-etc.html. 48 Using the most recent data for disasters that occurred much further in the past is, of course, problematic. Some disaster impact zones may have been more sparsely populated at the time of the disaster occurrence than they are now. Therefore, the likelihood of protest mobilization in such cases may be slightly lower than currently assumed. 39

Figure 2. Environmental disasters and associated protests: Europe, 1900–present.

Figure 3. Environmental disasters and associated protests: North America, 1900–present.

Figure 4. Environmental disasters and associated protests: Asia, 1900–present.

2.3 Discussion

Upon examining the maps, one can see slight regional differences in post-disaster protest patterns. Europe has experienced most large-scale disasters and most post-disaster protests with the highest number of participants. Of all disaster types, mine leaks have generated the fewest public protests in both North America and Europe.49 This is not surprising, given the discussion in Chapter 2 on mining accidents being generally viewed as less threatening. In contrast, oil spills have resulted in the highest number and variety of protests – some of them transnational. With respect to nuclear disasters, only those originating in North America and Japan led to protests, with the Fukushima disaster having the largest social impact in Japan and Europe. The proximity of disasters to population centers and protected areas does not seem to consistently affect post-disaster protest mobilization. I discuss these findings in detail below. As seen from the maps in Figures 2 to 4, mining disasters with large environmental impacts are often followed by minimal public response. Of four large North American mine leaks, small public protests erupted only in the aftermath of the Mount Polley disaster in Canada. Similarly, in Europe, out of four large mining disasters, only one – Talvivaara in Finland – led to protests. Population density in disaster zones did not seem to affect the protest occurrence in the expected direction, since mine leaks that occurred close to more densely populated areas (such as Los Frailes in Spain and Kingston in the USA) generated no public protests. Both Mount Polley and Talvivaara mines are located in remote and largely unpopulated areas, with Talvivaara being much further from population centers than Mount Polley. In both cases, protests occurred in cities (Vancouver for Mount Polley and Helsinki for Talvivaara); although, in Finland, there were also protests in smaller towns. The sizes of protests after these two disasters differed – in Vancouver, the protest was small, with only a handful of people, while in Finland, it ranged from ten to 20,000 protesters (Nuclear Heritage; Yle 2012). This suggests that population density of the impact zone may not be a deciding factor in the post-disaster protest size. A similar pattern emerges when mapping the proximity of these disasters to protected areas. Unlike in the USA, the European and Canadian mines can be found close to protected areas. Curiously, the two Nordic mines – Aitik and Talvivaara – are surrounded by national

49 There were no large-scale mine leaks and oil spills in Japan and South Korea (i.e., the only Asian OECD members).

43 parks, but these are much larger and much less scattered in Sweden than in Finland. Yet, while the Finnish protesters demanded the mine’s closure (Yle 2012), no public protests followed the Swedish disaster. In comparison, Mount Polley is located near large stretches of protected areas – the Tweedsmuir Provincial Park, Bowron Lake Provincial Park, Wells Gray Provincial Park, and several others. The associated protest occurred in Vancouver where the participants were primarily Vancouver-based indigenous activists. Calling the water their “lifeblood” and the salmon “the backbone of [their] communities,” these activists drew attention to the disaster’s impacts on the subsistence value of the damaged environment (Richmond 2014). Therefore, although the impact zone, as it relates to the proximity to protected areas, may be less important in mobilizing individuals directly affected by the disaster, it may have mobilizing potential in more distant communities. With respect to North American and European nuclear disasters, only the Three Mile Island (TMI) event generated protests (in the United States). These, again, occurred not within the disaster’s impact zone but in two densely populated areas: Washington, DC, and Rocky Flats, a nuclear weapons production facility north of Denver, Colorado. It is more likely that these two locations are significant for political rather than geospatial reasons (DC being the seat of the US government, and a nuclear weapons plant symbolizing a politically charged issue of nuclear weapons proliferation). TMI is relatively close to a large marine protected area, but the newspaper coverage of the post-disaster protests reveals that the protesters were part of an anti- nuclear movement – their main concerns were the safety of nuclear facilities across the country and the ethical issues surrounding nuclear weapons rather than environmental damage from the TMI disaster (Allen 1979; Lynton 1979; Martin 1979). The remaining nuclear disasters in North America and Europe, although close to densely populated zones and similarly-sized protected areas, generated no public protests. Again, this suggests that geospatial characteristics of large environmental industrial disasters may not be a deciding factor in post-disaster protest mobilization. The nuclear disasters in Japan support this conclusion. Two major nuclear disasters occurred in Japan within a decade – one at the Tokaimura plant in Tokai, Ibaraki in 1999, and the other at the Fukushima Daiichi plant in 2011. Both facilities are located close to densely populated zones and protected areas, but protests erupted only after Fukushima, and only in Tokyo (although, there was a petition in Japan after the Tokaimura disaster). Fukushima, however, inspired protests in other countries across the world –

44 the most notable due to their size were those in Germany. These were, of course, outside of the Fukushima impact zone, and occurred in densely populated areas, specifically in Berlin, Hamburg, Munich, and Cologne (see Appendix 1, Table 31). Lastly, oil spills have lined the coastlines of North America and Europe. Protests linked to these disasters vary the most out of all disaster types, in both location and size. The Deepwater Horizon and Exxon Valdez spills generated the largest number of protest events – across the United States and, in the case of Deepwater Horizon, beyond the US borders. Yet, it was the Prestige oil spill that was followed by the largest demonstrations (see Appendix 1, Table 31). The Prestige disaster occurred far away from population centers, with related protests taking place in densely populated areas of Spain. In contrast, the Independenta spill in Turkey, which led to no public protests, occurred in an area with a very high population density. Likewise, both the MT Haven and Amoco Cadiz spills were close to populated areas, but the latter was followed by protests while the former was not. Similar pattern is evident for other oil spills in Europe and North America. Being close to the disaster does not seem to motivate protests consistently, but once protests occur, they are likely to be in densely populated areas. A similar conclusion arises from the mapping of protected areas in both North America and Europe. The Exxon Valdez disaster is the only oil spill clearly linked to concerns over environmental damages. In Europe, even though most of the spills occurred near marine protected areas, only the Sea Empress, Amoco Cadiz, Aegean Sea, and Prestige disasters generated protests. Therefore, the findings from this preliminary GIS analysis fully support neither hypothesis 1 nor hypothesis 2.50 Hypothesis 3, however, seems to stand. The only truly remote disaster in this dataset is the oil spills from Odyssey– no protests are linked to this event.51 The Andros Patria and Betelgeuse disasters, followed by no protests, also occurred in remote locations (although these were relatively closer to protected areas). Overall, this suggests that sudden grievances, motivated by the disaster damage of areas that people value, may be a necessary but not sufficient driver of post-disaster protest mobilization.

50 Although, the sizes of protests that did occur do align with expectations of hypothesis 1. This GIS analysis could be improved through autocorrelation – this means using ArcGIS to discern correlation as opposed to doing it visually from observing maps. Autocorrelation can be used to decide whether there is a pattern among the mapped events/entities or whether their distribution is random. ArcGIS can be used to conduct quantitative tests of such clustering (see Gimond 2018). 51 Arguably, more cases are needed to make this claim convincingly. 45

More specific theoretical propositions arise from this analysis, and particularly from examining Figures 2 to 4. Population density, for example, may matter in post-disaster protests in at least two ways. First, if the disaster impact zone is sparsely populated, protests may not develop simply because the pool of potential protesters is not large enough. In the immediate disaster aftermath, directly affected individuals may be more concerned with disaster response and recovery than with mobilizing for protest. Second, outside of the disaster impact zone, protests tend to occur in densely populated areas. Aside from widening the pool of potential protesters, such places may house individuals who not only hold strong environmental values but, by not having to participate in clean-up, are also available for protest. A disaster is likely to encourage protest mobilization of directly unaffected groups if the risk of occurrence of a similar event is either sufficiently high or perceived as such (Birkland 1998, Bishop 2014). Since this risk threshold is rather subjective, framing (discussed in Chapter 5) is likely to play a crucial role in shaping disaster risk perception as well as public reaction to disaster impacts. Furthermore, this mapping of disasters and protest suggests that intrinsic values of nature are unlikely to motivate people to take action, since no protests have been linked to natural areas with no clear value to humans. The analysis does not offer clear support for the motivating role of instrumental value. It does, however, suggest that, at the minimum, we should re-visit the concept of grievance. Psychological studies claim that we are more risk taking in face of losses (for example, Tversky and Kahneman 1981), but some losses seem more important than others. Sudden grievances may not be simply about a perception of loss but a perception of specific type of loss, some perhaps based on instrumental values, while others on something else.

Conclusion

The aim of this chapter was to assess whether and how disaster damage influences post- disaster protests. Since industrial disasters harm the environment valued by humans, they generate sudden grievances that motivate protest. Put simply, these grievances are felt because of what they affect. A geospatial analysis was therefore used to assess the relationship between industrial disasters, their proximity to locations that people value, and post-disaster protests. The findings reveal that disaster damage is likely to be important in generating sudden grievances but is not sufficient for post-disaster protest mobilization. Perhaps not surprisingly, protests are more

46 likely to spring up in densely populated areas, and less likely after mining disasters. Oil spills, however, tend to motivate people more than nuclear disasters, a finding that goes contrary to theoretical expectations. Disaster proximity to protected and populated areas seems to serve as a motivational factor, and no protests have been linked to areas that are of little instrumental value to humans. This gives some support to the motivational effect of instrumental rather than intrinsic value of nature, and poses a further research question: While people have been found to value nature intrinsically (Vucetich, Bruskotter, and Nelson 2015), why do such attitudes fail to motivate political action? To answer such question, we might have to move away from the traditional understanding of grievance (i.e., based on severity) and instead focus on the meaning of grievances through consideration of both material and ideational factors (see Simmons 2014). For example, an environmental disaster may represent ideas of technology, progress, or broader forces of environmental destruction, depending on place and time. An examination of how and why individuals value certain parts of nature might shed more light on the relevance of grievances in protest mobilization and explain why there is such mixed empirical evidence when it comes to their effects. As noted in the introduction, this chapter was not meant to test hypotheses but instead search for patterns between disaster damage and protest in geospatial terms. Further work is needed to build on the preliminary findings. There are two common ways of testing hypotheses through GIS: comparing observed patterns to those generated by an independent random process (IRP),52 and the Monte Carlo technique, which involves formulating competing hypotheses, simulating many realizations of the hypothesized process, and comparing observed data to the simulated patterns (see Gimond 2018). The preliminary findings from this analysis offer three additional ideas for further GIS work: mapping of post-disaster protests in relation to nuclear power plant facilities, an examination of the role of disaster recurrence and cumulative impacts, and an analysis of the relationship between post-disaster protests and local dependence on the responsible industry. First, the location of post-Fukushima protests prompted a question of whether some significant geospatial factor – other than population density – has affected the protest mobilization in Germany. Given the public fear of nuclear energy production, the density and

52 The ANN tool in ArcMap can be used to test whether the observed pattern is from a random process.

47 size of nuclear power plant facilities may have been one such factor. To probe this assumption, preliminary mapping53 has been conducted for Germany and Asia (see Figures 5 and 6). Together, Figures 5 and 6 show the extent of the Fukushima effect in Germany. Completing the mapping of nuclear reactors in the rest of the countries with major nuclear disasters (the USA, Canada, the United Kingdom, Switzerland, and France) would allow for potential patterns to emerge. Second, certain geographic regions or industrial zones (e.g., oil tanker routes or mining towns) have a higher chance of disaster occurrence than others.54 Theoretically, in such areas, there are two potential opposing effects of disaster recurrence on protest. Frequent recurrence may inspire better disaster preparedness or may come with a lower ‘shock value.’ Over time, this may lead to desensitization, and therefore less public outrage and lower likelihood of post- disaster protests. Conversely, frequent disaster recurrence is likely to result in cumulative destructive impacts on the environment, which may prompt the public to demand better environmental protection and better disaster preparedness. Aside from GIS, a process tracing study could shed light on the validity of these theoretical propositions. The third area of further study involves mapping local dependence on the industry that caused the disaster. In a disaster impact area, individuals affected by pollution may demonstrate greater support for environmental protection – because of the direct impact that pollution has on their lives. Alternatively, even when faced with environmental costs, they may maintain their support for the local industry in fear of losing employment or other benefits the industry provides.55 Such individuals and groups would be therefore less likely to participate in a protest aimed against that industry. They may, however, engage in a post-disaster protest against other actors, including various levels of government. By allowing for someone to be blamed, industrial disasters open different pathways to potential social clashes. ‘Acts of God’ may bring people together to endure their faith, but ‘acts of corporations’ bring them together to demand change.

53 The nuclear reactor data are from the World Nuclear Association’s Reactor Database, available at http://www.world-nuclear.org/information-library/facts-and-figures/reactor-database.aspx. The data were retrieved on December 30, 2017, and were modified for import into Excel using TextMate to reformat. Coordinate information was retrieved from Google Maps using latlong.net. 54 For example, there were three large oil tanker disasters in Spain along the same stretch of coastline between 1976 and 2002, with the Prestige oil spill being the last of them (and the only one followed by large-sized protests). 55 Such support would likely be stronger in communities affected by nuclear and mining disasters (because of the ‘stationary’ nature of the industry). Oil tanker disasters are less likely to see local support for the oil industry in the impact zone unless the industry happens to be particularly well established there.

The next chapter explores some underlying social conditions under which such demands are likely to develop.

Figure 5. Nuclear power plant facilities and protest before the Fukushima disaster.

Figure 6. Nuclear power plant facilities and protest after the Fukushima disaster.

Chapter 4. Structural Conditions for Post-Disaster Protest

Many scholars have come to an agreement that in order to understand the emergence of social movements and protests, we need to examine the underlying social, cultural, political, economic and other processes that define and shape our societies. Factors such as government regime type, society’s levels of income, and ethnic or religious cleavages are often linked to the onset of uprisings and revolutions, or to other types of conflict (see, for example, Pop-Eleches and Robertson 2015). Similarly, the existence and character of civil society, political culture, and the level of socio-economic development have been identified as some of the vital elements needed for the emergence of social movements and non-violent resistance (Opp 2009; Chenoweth and Ulfelder 2017). How critical are structural conditions in non-violent protest after environmental disasters? What type of social context is most conducive to the development of post-disaster protest movements? As a whole, the existing studies of environmental disasters do not systematically investigate the role of structural conditions in post-disaster protest. As the scale and urgency of environmental degradation grows, environmental protests have been increasing in numbers and intensity, reflecting a growing popular discontent with the state of the environment. In the past, such protests have led to significant milestones in the environmental movement – the establishment of Earth Day, more stringent regulations for environmental protection, as well as a shift in environmental culture and public perceptions of the environment from conservation to the responsibility to protect. Industrial disasters such as the 1969 Santa Barbara and 1989 Exxon Valdez oil spills were at the core of many of these changes (Molotch 1970; Hoffman and Jennings 2010). The potential of these events to generate such major shifts in societies justifies a closer examination of their interactions with relevant socio-economic and political conditions. The goal of this chapter is to better understand what structural conditions are conducive to post-disaster protest emergence. These factors are not treated as ‘causes’ of protest but viewed as enablers. This is because many of them are static or slow-changing (e.g., industrialization, post-Cold War environment) and on their own cannot explain protest emergence. Unlike prevailing studies, I evaluate both economic and environmental grievances, in addition to other factors theorized to facilitate protest: resources, political opportunity structure, and modernization. Through a qualitative comparative analysis (QCA), I examine the economic and

52 socio-political contexts in thirty-nine cases of environmental industrial disasters, searching for both necessary and sufficient conditions for the emergence of post-disaster protest movements. The findings reveal that only one necessary condition is required for the emergence of nonviolent protest after an environmental disaster: absence of state repression. While this is not surprising, the fact that other factors – an especially established movements – are not necessary for protest contradicts some prevailing literature. No single condition is sufficient, on its own, for the emergence or absence of post-disaster protest. However, I find several combinations of individual conditions that are sufficient for both protest emergence and protest absence. These are different configurations of structural factors such as economic conditions, degree of state repression, and the existence and strength of an established environmental movement and its resources. Many of these combinations of factors support the findings of past studies, although I am able to illustrate with more clarity how the factors work together to foster protest. One finding that differs from past work relates to the role of inequality in predicting protest. I find that it is not the expected high inequality but the lack of it that together with other factors aid post-disaster protest emergence. The findings also reveal that the structural conditions examined here are neither sufficient nor necessary for determining any particular protest size. This confirms that different factors account for protest emergence versus whether and to what extent the protest grows. I examine factors that likely influence protest size in the next chapter. This chapter proceeds with a brief discussion of the prevailing literature on structural conditions and environmental disasters. I then outline the chapter’s method of analysis (i.e., QCA), specifying the main steps, theoretical expectations, data collection, coding, analysis, and findings. The chapter concludes with a summary and suggestions for future research.

1. Structural Conditions, Disasters, and Protest

Disasters can facilitate political openings and subsequent organizational, institutional or policy changes (Olson and Gawronski 1985; Birkland 1997 and 1998; Busenberg 200; Olson 2008; Birkmann et al. 2010; Pelling and Dill 2010). Such effects of disasters, however, are likely contingent upon some pre-existing political, economic, and social conditions. Studies of the effects of natural disasters on violent conflict generally focus on economic conditions such as the level of a country’s development and wealth, income inequality, and resource scarcity (Drury

53 and Olson 1998; Brancati 2007; Nel and Righarts 2008; Bearsley and McQuinn 2009). These studies, however, do not provide insights into the emergence of nonviolent campaigns. Such campaigns and the processes through which they develop are distinct from violent rebellions, and the structural conditions that can explain the latter do not fare well in providing insights into the former (Chenoweth and Lewis 2013; Chenoweth and Ulfelder 2017). Perhaps because of their diverse research focus, industrial disaster scholars offer no decisive set of structural conditions that enable non-violent protest mobilization after environmental disasters. Some study the role of economic factors on public opinion such as local economic benefits in the disaster impact area (Bishop 2014; see also Hoffman and Jennings 2010, p. 13). Others have investigated the mobilizing effects of social structures at the community level. For example, Harvey Molotch (1970) has argued that the upper and upper middle-class residents of Santa Barbara were a crucial element in the public response to the 1969 Santa Barbara oil spill. They were “a large number of worldly, rich, well-educated persons— individuals with resources, spare time, and contacts with national and international elites— [who] found themselves with a commonly shared disagreeable situation: the pollution of their otherwise near-perfect environment” (p. 131). In addition, these residents were “literature [sic] and leisured”; they had the ability to “read, to ponder and to get upset” (Molotch 1970, p. 143). Studying the Deepwater Horizon oil spill, Hoffman and Jennings (2010) present a wider range of structural factors, including cultural, political, and identity-related, that are likely to facilitate institutional changes in environmental management and fossil fuel production (but not necessarily protest mobilization). Yet others emphasize the enabling role of the pre-existing environmental movement and the existence of social networks in the likelihood of sustained protest mobilization (Walsh 1981; Fernandez and Pena 2004; Lendon and Martin 2007; Hasegawa 2014). Since the prevailing disaster studies offer little systematic guidance on which structural conditions are comparatively more or less likely to enable post-disaster protest mobilization, I primarily draw on the existing literature of nonviolent resistance, as discussed below.

2. Examining Post-Disaster Protest through QCA

This chapter evaluates the salience of structural factors in post-disaster protest through the use of Qualitative Comparative Analysis (QCA). QCA is a desirable method when the

54 phenomenon under study is marked by causal complexity. It is best applicable for mid-size N, and often combined with a within-case analysis to evaluate statements about sufficient and/or necessary conditions (see Schneider and Wagemann 2010; Wagemann and Schneider 2010). Given the smaller number of cases under study, QCA is more suitable for this research than a large-N analysis. QCA requires collection of raw data, calibration of set conditions, and coding of the data based on these conditions. For this research, the raw data were collected from multiple sources, as outlined below, and include quantitative variables and secondary literature.

2.1 Theory and raw data

In general, two broad types of factors may aid the emergence of nonviolent campaigns: agency and structure. Agency captures the skills and methods of activists who exploit structural conditions to achieve their goals (see Sharp 2005; Ackerman 2007). Their activities rely on “the cultural stock of how to protest and how to organize,” and include the framing of injustice and political goals, fundraising, running an office, as well as “repertoires of contention” (e.g., organizing demonstrations) (Zald 1996, p. 267). Structure refers to the underlying societal conditions that enable or constrain activists’ activities. Three of these have been found to make nonviolent protest likely to occur: grievances, resources, and political opportunities. While Chapter 3 focused on sudden grievances, here, grievances as structural conditions refer to long-term underlying roots of public discontent. A disaster may disproportionally impact certain groups relative to others (Drury and Olson 1998; Adeola 2011). Poor and marginalized groups affected by an environmental disaster may therefore have a mix of underlying grievances (e.g., economic, social or environmental) that the event exacerbates. In rich industrialized democracies, two types of grievances are likely to influence protest emergence after environmental disasters: economic and environmental. First, poor economic conditions – specifically slow economic growth, high inflation, high income inequality, and high poverty rate – are likely to lead to public frustration and a sense of shared grievance against the wealthy minority, and therefore higher likelihood of protest (Gurr 1970; Cederman, Weidmann, and Gleditsch 2011). Economic grievances can be captured by economic indicators, and proxied by

55 annual GDP growth rate, the consumer price index, the Gini coefficient, and the infant mortality rate (World Bank 2016).56 Second, environmental grievances are likely to stem from the dominant culture as it relates to the environment. Such culture – beliefs, norms and institutions – determines how major environmental disasters are viewed at the time (Hoffman and Jennings 2010; see also Hulme 2009). In general, environmental culture that is attentive to environmental issues and is supported by state-led environmental protection is likely to generate fewer environmental grievances (Weersink and Raymond 2007; Hoffman and Jennings 2010). In Western countries, major shifts in environmental culture occurred with the emergence of environmental state ministries or agencies. Specifically, the establishment of the US Environmental Protection Agency (EPA) in 1970 marked the beginning of institutionalization of new norms of environmental protection, which rippled across the Western world (Schreurs 2002, p. 35; Hoffman and Jennings 2010). The existence of environmental culture can therefore be proxied by the existence of state-led environmental institutions.57 The second set of structural conditions likely to influence the emergence of protest is resource related. Resources needed for protest mobilization can be human, financial or informational. They are determined by the number of potential protesters available as well as activists’ organizational capacity and skills (McCarthy and Zald 1973, 1977; Zald and McCarthy 1979). Chenoweth and Ulfelder (2017) identify several factors that shape the availability of resources for protest mobilization. Among these are urbanization,58 the “carrying capacity”, and civil society’s organizational capacity and learning.59 Urbanization facilitates protest because of the ease of recruitment – higher concentration of the population enables coordination and cooperation (see also Gurr 1970; Goldstone 1991; Wallace 2013). This concept can be proxied by country’s percentage of urban population (World Bank 2016; UN DESA 201860). The

56 The data are available from the World Bank’s World Development Indicators at https://data.worldbank.org/products/wdi. 57 Environmental culture can also be proxied by the stringency of environmental regulations, which can be measured through various indices such as the Environmental Performance Index (EPI) published by the Yale University and Columbia University. However, these indices are not suitable for this research, because they are relatively new and suffer from lack of older data. 58 The GIS analysis in Chapter 3 also points to the significance of urbanization in protest emergence. 59 Chenoweth and Ulfelder (2017) also include youth bulge (because youth is easier to recruit), regional contagion (i.e., from other movements in the region), and civil war (which makes participation in campaigns riskier). No data on youth bulge were available for OECD countries, so I excluded this variable. Regional contagion is not a structural condition per se but a triggering event (or a series of triggers). Lastly, there were no civil wars in the cases under this study, and all other campaigns are already captured by the concept of “carrying capacity”. 60 The data are available at https://esa.un.org/unpd/wup/Country-Profiles/.

56 carrying capacity refers to the number of available protesters – since the supply of potential protesters is limited, the public cannot be otherwise engaged in other campaigns. The carrying capacity is given by the existence of any ongoing campaign in a country and can be assessed using the Nonviolent and Violent Campaigns and Outcomes (NAVCO) dataset published by the University of Denver (Chenoweth and Lewis 2013). High carrying capacity is linked to higher likelihood of protest. Lastly, civil society’s organizational capacity and learning determine the movement’s strength. The structure of domestic civil society (and specifically the existence of established environmental or anti-industry movements) prior to a disaster is likely to encourage post-disaster protest mobilization. This is in part due to pre-existing resources such as established organizations with staff and legitimacy to express opinions on the issue at hand. Strong established movements (i.e., those with some degree of institutionalization and sufficient resources) are likely to mobilize sympathizers easier and in larger numbers than smaller movements (McAdam et al. 1996; Tilly 2004; Saunders 2013). Movement’s strength can be measured along at least three dimensions: the movement’s organizational base and resources, alliances, and campaigns. These can be operationalized in several ways, as shown in Table 7. These dimensions also, to some extent, capture what Charles Tilly (2004, p. 4) believed constitutes movement strength: a combination of worthiness, unity, numbers, and commitment.

Table 7. Measuring social movement strength.

Dimension Indicators/Metrics

Organizational base - Existence of SMOs – organizational headquarter and offices, staff; and resources - Size of budget and steady revenue stream; Alliance building - Number, composition, and diversity of partnering groups; - Scale of alliance reach: regional, national, international; Campaigns - The number of strikes, riots and demonstrations; - The reach of a campaign (the number and variety of media covering the issue, the number of op-eds and articles published); - The number of reports, briefs, articles, and research tools produced

The third set of structural conditions relates to the political opportunity structures and the associated beliefs about the utility of protest action. The likelihood of protest increases when the costs are low, and the probability of success is high (Eckstein 2001; Tarrow 2011). Specifically,

57 a high degree of state repression (such as massive violence against protestors) increases the cost of protest, and therefore decreases its likelihood of occurring (Carey 2006; Davenport 2007). Following Chenoweth and Ulfelder (2017), I used the Cingranelli–Richards (CIRI) index on annual state practices regarding physical integrity rights to measure the degree of state repression.61 In addition, two types of broader structural changes may increase the likelihood of protest emergence (through affecting protesters’ beliefs about the utility of protest): the end of the Cold War, and modernization. First, the post-Cold War environment may be conducive to nonviolent mobilization, given the global support for peaceful democratization processes (Bunce and Wolchik 2011; Chenoweth and Stephan 2011). Second, economic development has been theorized to lead to urbanization, improvements in public health and literacy, increased communication, and shifting values, with the emerging middle class being more prone to protest mobilization as they demand greater rights (Lipset 1959; Inglehart and Welzel 2005, p. 134). Factors linked to modernization include industrialization, communication, and trade liberalization (Chenoweth and Ulfelder 2017).62 These can be proxied by the manufacturing and services as percentage of GDP, mobile phone subscriptions rate, and membership in GATT/WTO, respectively.

2.2 Set calibration and coding

Qualitative comparative analysis is concerned with relations between sets and uses ‘the language’ of sets rather than the language of variables. It is a type of empirical comparative approach, which categorizes data according to set membership scores (e.g., crisp, fuzzy, multi- value). In this chapter, I use a crisp-set QCA, where only dichotomous sets are processed. In other words, a case can either be a full member (1) or full non-member (0) of the set. This establishes qualitative (not quantitative) difference between cases. QCA uses ‘conditions’ (as opposed to independent variables) and ‘outcomes’ (as opposed to dependent variables) (see Schneider and Wagemann 2010, p. 404).

61 CIRI is available at http://www.humanrightsdata.com/p/data-documentation.html. 62 Chenoweth and Ulfelder (2017) also include education, proxied by secondary school enrollment rate. The data for OECD countries were not available for this indicator. As such, it has been excluded from this analysis.

In this chapter, I am interested in finding out under what conditions post-disaster protest movements occur in democracies. My universe of cases consists of 38 environmental disaster events – 21 oil spills, eight mine leaks, and nine nuclear disasters that occurred between 1900 and 2014 in rich industrialized democracies (see Chapter 2, Tables 2 to 4). Both the outcome and the conditions are understood in terms of membership in particular sets. In this research, the membership in the crisp set of post-disaster protests is the ‘outcome’. As noted in earlier chapters, the process of determining whether a post-disaster protest occurred (and if so, how large it was) entailed a survey of both the newspaper articles and scholarly literature on the respective disasters. The sample includes both members (i.e., protest occurred, coded as 1) and non-members of this set (i.e., protest did not occur, coded as 0).63 Furthermore, there are 13 conditions, as shown in Table 8 below.

Table 8. QCA conditions.

1. set of countries with economic growth, 2. set of countries with high inflation, 3. set of countries with high income inequality, 4. set of countries with large poverty, 5. set of countries with institutionalized (i.e., state-led) environmental culture, 6. set of countries with high urbanization, 7. set of countries with several concurrent ongoing campaigns, 8. set of countries with strong environmental or anti-industry movements, 9. set of countries with a high degree of state repression, 10. set of countries existing after the Cold War, 11. set of countries with high industrialization, 12. set of countries with a high rate of communication, 13. set of countries with trade liberalization

I transformed the qualitative and quantitative information on each case into set membership scores through set calibration and respective coding of each case. The process of set calibration requires establishment of qualitative anchors that allow assignment of set membership scores (for crisp sets, scores of 0 or 1) based on some prior theoretical knowledge and a general understanding of the cases (Ragin 2008; Schneider and Wagemann 2010). Calibration criteria serve to assess how well the cases meet the membership requirements (i.e., deciding what it takes for a country to be considered democratic) – in other words, measurement

63 Similarly, I have coded protest size as small, medium, and large (based on the criteria identified in Chapter 3) and assigned membership scores based on occurrence.

59 requires interpretation (Ragin 2008). I briefly describe the calibration threshold of my thirteen conditions below. The first three conditions were sets of countries with economic growth, high inflation, high income inequality, and high poverty rate. Membership in a set was coded as 1, and non- membership as 0. The threshold for economic growth was established based on the negative socio-economic impacts of stagnation, which is a period of low or no growth, generally less than 2 to 3 percent. This threshold was selected, because lack of economic growth is expected to generate grievances, and stagnation produces effects such as high unemployment that are in line with such expectations. The raw data were coded as follows. In the year when disaster occurred, the annual value of GDP growth per capita more than 2 percent qualified the state for membership in the set of countries with economic growth (coded as 1). Similarly, a country with high inflation in the disaster year was a member of the respective set (coded as 1), whereas if the country experienced low or moderate inflation, it was a non-member of the set (coded as 0). The threshold is based on the available literature, and specifically, on the understanding of the negative socio-economic effects of inflation. Omay (2010) found that significant negative effects of inflation manifest at 2.52 percent (measured by percent change in the Consumer Price Index). Income inequality, measured by Gini coefficient, ranges from 0 (complete equality) to 100 (one household possesses all the nation’s income). Among the OECD members, the highest levels of inequality are in the 40s, and the lowest in the 20s (OECD 2018). Since the variation within this group of interest is less pronounced than the variation between developed and less developed countries, the threshold was set as the value of Gini of 35 to allow for better comparison among OECD members. Countries with inequality above this threshold were coded as members of the set of countries with high inequality. Furthermore, inequality was measured over a five-year period (with the fifth year being the disaster event year) to account for the slow- moving nature of its grievance-generating effects. Infant mortality rate was measured as infant death per 1,000 live births. The OECD data reveal that 40 deaths per 1,000 births is the upper bound among OECD members (OECD 2018b). Again, to allow for comparison within this group, the calibration threshold was set accordingly, with 21-40 deaths coded as high infant mortality rate. This qualified the state in a given disaster year as a member of the set of countries with high poverty level. The last of the grievance conditions – institutionalized environmental culture – was captured by the existence of national

60 environmental agencies or ministries, where members (i.e., countries with such institutions) were coded as 1 and non-members (i.e., countries without such institutions) as 0. Three resource-related conditions are likely to encourage protest emergence: sets of countries with high degree of urbanization, high protest carrying capacity, and strong environmental or anti-industry movements. Urbanization was measured as the percentage of urban population out of total population of the country in a given year. The calibration threshold was set at mid-point, where 51 to 100 percent was needed for membership in the ‘high urbanization’ set. For the carrying capacity, membership was defined by the occurrence of other campaigns in the country in the disaster year.64 Civil society’s organizational capacity and learning can be captured by two sets: a set of countries with environmental or anti-industry movement existing in a given disaster year, and a set of countries where such movements are strong. Because of the multi-dimensional nature of ‘strong’ movements and due to lack of available data, I have decided to focus only on organizations with international alliances. The evaluation of the membership was case-specific and based on a survey of secondary literature (see Table 9 for sources). The conditions linked to political opportunity structures are: set of countries with state repression, countries that exist in the post-Cold War period as well as countries with trade liberalization, high industrialization, and high level of communication. The CIRI dataset was used to create the threshold for membership in the repression set. The dataset is composed of two indices: physical integrity rights index, and empowerment rights index. The former includes country’s record of torture, extrajudicial killings, political imprisonment, and disappearances. It ranges from 0 (no government respect for these rights) to 8 (full government respect for these rights). The latter considers countries’ freedom of movement, freedom of speech, workers’ rights, political participation, and freedom of religion. It ranges from 0 to 10, with the newer version ranging from 0 to 14.65 I have coded the raw data based on a mid-point threshold, considering a five-year period prior to (and including) the disaster year to account for the slow- moving nature of the associated grievances.

64 The data came from the NAVCO dataset v2.0, available at https://www.du.edu/korbel/sie/research/chenow_ navco_data.html. 65 The newer version of this index (since 2007) includes foreign movement, domestic movement, freedom of speech, freedom of assembly and association, workers’ rights, electoral self-determination, and freedom of religion indicators. 61

The threshold for the membership in the post-Cold War set was the year 1991 (i.e., countries linked to disaster years after 1991 were coded as members). Similarly, membership in the set of countries with trade liberalization was based on country’s membership in the GATT/WTO, with members coded as 1 and non-members coded as 0. The last two sets are linked to modernization. The raw data for industrialization combined country’s value added (as a percentage of GDP) for manufacturing and services. Value added captures the contribution of different sectors to production (i.e., overall GDP), and is an established indicator of industrialization (UNIDO 2015). In advanced industrialized countries, the combined manufacturing and services value added is above 80% (Santacreu and Zhu 2018). Therefore, to allow for better comparison, the threshold for this set’s membership was set at 80%. Only countries in a given disaster year with value added of 81% and above were coded as members of this set. Lastly, the raw data on communication were comprised of rate of mobile phone subscriptions (per 100 people). The threshold was set at mid-point, with 51 to 100 subscriptions qualifying for membership in the set of countries with high level of communication. The thresholds for all conditions are summarized in Table 9 below.

Table 9. Concepts, indicators, and set membership thresholds.

Concept Indicator Source Set membership Expected effect threshold (1) on protest Economic growth Year to year percentage World Bank Below 2% - change in real GDP per capita

Inflation Consumer price index World Bank Above 2.5% + Income inequality Gini World Bank; Above 35 + Roser and Ortiz-Ospina (2016)

Poverty Infant mortality rate (per 1000 World Bank Above 20 deaths + deaths)

Environmental Existence of national Author’s Presence in a given - culture environmental coding year agency/ministry

Urbanization Percentage of urban World Bank; Above 50% + population UN DESA

Civil society’s Existence of movement Author’s Presence in a given + organizational coding66 year capacity and learning Alliance building Member of + international networks

Carrying capacity Any ongoing campaign in NAVCO Presence in a given + country dataset v2.0 year

State repression CIRI’s physical integrity index CIRI Above mid-point -

Post-Cold War Year is after 1991 Author’s Presence after the + period coding year 1991

Industrialization Manufacturing and services as World Bank Above 80% + percentage of GDP

Communication Mobile phone subscriptions World Bank Above 50 + per 100 people subscriptions

Trade liberalization GATT/WTO member Author’s Membership in the + coding organization

66 McCormick 1991; Kamieniecki 1993; Diani 1995; Konttinen et al. 1999; Nave 2000; Schreurs 2002; Badruddin 2003, p.110; Fillieule 2003; Garner 2004, p. 49; Mehta 2004, p.38; Johnson and McCarthy 2005; Rootes 2005; Rootes 2006; Jimenez 2007; Leonard 2007; Buckingham 2008, p.41; Leonard 2008, p. 215; Vyas 2010; MacDowell 2012; Hasegawa 2014; Curran 2015; Sahin 2015; Hummel 2016; Queiros 2016; Swedish Environmental Protection Agency; Wiklund.

2.3 Analysis and discussion

QCA analysis is generally carried out in four steps: searching for necessary conditions, representing empirical evidence in a truth table, identifying sufficient conditions by logically minimizing the truth table, and doing within-case analyses in typical and deviant cases (Ragin 2008; Schneider and Wagemann 2012). In this analysis, all calculations were implemented with a QCA package and set methods package in R (Dusa 2007; Medzihorsky et al. 2016). The first step was to determine whether any of the conditions are necessary for the outcome. In QCA, necessary conditions mean that “instances of an outcome constitute a subset of instances of a cause” (Ragin 2008, p. 44), and therefore no case may contain the outcome without the condition (Schneider and Wagemann 2012).67 Two factors are linked to the evaluation of necessity in QCA: consistency and coverage. Consistency refers to “the proportion of cases with a given cause or combination of causes that also display the outcome” (Ragin 2008, p. 46). A hypothesized subset relation that is not consistent suggests the theory is not supported. Consistency scores should be as close to perfect consistency (1.0) as possible, and generally no lower than 0.9 (Ragin 2008; Schneider and Wagemann 2012). Coverage denotes the empirical relevance of the necessary condition, with high score values indicating relevance and low values indicating trivialness of the condition (i.e., when the outcome set is much smaller than the condition set).68 For necessary conditions, coverage score of less than 0.5 are rare, and suggest the necessary condition is empirically irrelevant. I find seven conditions with high consistency values: environmental culture, urbanization, established movement, movement with international allies, post-Cold War environment, industrialization, and economic liberalization (see Table 10 below).69 None of these, however has a high enough coverage score, which suggests these conditions may be necessary but trivial (i.e., the conditions cover barely 50% of the total membership of the outcome). The negations of the conditions yield similar results except for state repression – lack of this condition has both a high consistency and coverage value, suggesting that absence of state repression is a necessary condition for protest to emerge. While this is in line with the prevailing literature and theoretical

67 In other words, “if X is necessary for Y, then X is a superset of Y, whereas if X is sufficient for Y, then it is a subset of Y” (Schneider and Wagemann 2012, p. 139). 68 This means that trivial necessary conditions are those that are strongly present in most cases regardless of the cases displaying the outcome (Ragin 2008; Schneider and Wagemann 2012). 69 I had to narrow the original universe of cases, because the QCA software is unable to work with conditions and cases with missing values. This reduced my universe of cases from 38 to 16 disaster events. This also means that there are no pre-1992 cases available for this analysis.

64 expectations, the other results represent important empirical findings that go against the tendency in some literature to consider variables such as established environmental movement to be preconditions for related protest (Walsh 1981; Lendon and Martin 2007; Hasegawa 2014). This analysis also suggests, in line with some literature, that grievances are not necessary conditions for protest (McCarthy and Zald 1973, 1977; Zald and McCarthy 1979; McAdam et al. 1996; Rucht 1996). Of the three economic indicators, only the absence of growth approaches (but does not meet) the required threshold for consistency score. Curiously, environmental culture has a consistency score of about 0.9, meeting the requirement for necessity. The coverage, however, is low, suggesting that environmental culture might be a trivial necessary condition for protest emergence. This goes contrary to the theoretical expectations that existence of such culture lessens public grievances and therefore lowers the likelihood of protest. The difference between economic and environmental grievances suggests that a closer examination of the role of different types of grievances in protest emergence might be warranted. This conclusion is in line with the findings and conclusions presented in Chapter 3.

Table 10. Analysis of necessary conditions for the outcome ‘protest’.

Condition Consistency Coverage ~ Condition Consistency Coverage

GROWTH 0.250 0.286 not GROWTH 0.750 0.600 INFL 0.500 0.444 not INFL 0.500 0.500 INEQ 0.250 0.333 not INEQ 0.750 0.545 POV 0.000 NA not POV 1.000 0.471 ENV 0.875 0.438 not ENV 0.125 1.000 URBAN 1.000 0.500 not URBAN 0.000 0.000 MOVM 1.000 0.471 not MOVM 0.000 NA IALLY 1.000 0.500 not IALLY 0.000 0.000 OCAMP 0.375 0.600 not OCAMP 0.625 0.417 REPR 0.000 0.000 not REPR 1.000 0.571 POSTCW 1.000 0.471 not POSTCW 0.000 0.000 INDUSTR 1.000 0.500 not INDUSTR 0.000 0.000 COMM 0.750 0.667 not COMM 0.250 0.250 TRADE 1.000 0.471 not TRADE 0.000 NA

An analysis of different protest sizes as outcomes yields high consistency values across almost all conditions but very small coverage values, suggesting these conditions are trivial for the protest size (see Appendix 2, Tables 32 to 34). This suggests that while emergence of protest itself is linked to structural conditions, factors other than structural conditions determine the protest size. The second step in QCA analysis is to identify sufficient conditions, using a truth table. Truth tables are used to examine the distribution of cases across all logically possible combinations, given the sets of causal conditions. They assign each case to one of the 2k70 logically possible combinations of conditions (represented as the rows in the truth table) (Ragin 2008, p. 129). The goal is to “identify explicit connections between combinations of causal conditions and outcomes” (Ragin 2008, p. 125). Table 11 presents a shortened truth table for the outcome ‘protest’, with all logical types without empirical cases removed. The 17 cases are distributed over 13 logically possible combinations of conditions. A minimal consistency score for sufficient conditions is 0.75 (Ragin 2008, p. 136). As seen in the truth table, only the first four rows contain consistency values above this threshold. To reduce the complexity of results, I derive the ‘conservative solution’.71 As seen in Table 12, the solution term identifies three pathways with sufficient combinations of conditions for the outcome. The solution’s consistency is 1.000, which means that the degree to which the empirical observations are in line with the theorized set relations is high and significant. The coverage of the solution is 0.625 – almost 63 percent of the empirical cases that exhibit the set-theoretical relations of the solution also exhibit the outcome.

70 K is the number of conditions. 71 The generation of conservative solution is guided by the empirical information at hand (Schneider and Wagemann 2012, p. 162).

Table 11. Truth table for the outcome ‘protest’.72

GROW INFL INEQ ENV URB MOV IALLY OCAMP REPR POSTCW 0 0 0 1 1 1 1 0 0 1 0 1 0 0 1 1 1 1 0 1 0 1 0 1 1 1 1 0 0 1 0 1 0 1 1 1 1 1 0 1 0 0 1 1 1 1 1 0 0 1 1 0 0 1 1 1 1 0 0 1 1 1 1 1 1 1 1 1 0 1 0 0 0 1 1 1 1 0 0 1 0 1 0 1 1 1 0 0 0 1 0 1 1 1 1 1 1 0 1 1 1 0 0 1 1 1 1 1 1 1 1 1 0 1 0 1 1 0 1 1 1 1 1 1 1 1 1 0 0 1

INDUS COMM TRADE OUT N Cons Cases

1 1 1 1 2 1.00 1,3

1 0 1 1 1 1.00 16

1 1 1 1 1 1.00 2

1 1 1 1 1 1.00 8

1 1 1 0 2 0.50 5,6

1 1 1 0 2 0.50 4,10

1 0 1 0 2 0.50 14,15

1 0 1 0 1 0.00 12

1 0 1 0 1 0.00 17

1 1 1 0 1 0.00 7

1 0 1 0 1 0.00 13

0 0 1 0 1 0.00 11

1 0 1 0 1 0.00 9 Note: OUT = outcome value, N = number of cases in configuration, Cons. = consistency/sufficiency inclusion score.

72 The condition ‘poverty’ is not shown in the table, because all values in the column were zero.

Table 12. Conservative solution for the outcome ‘protest’.73

Configurations Cons. Raw coverage Unique coverage growth*INFL*ineq*env*URBAN*MOVM* 1.000 0.125 0.125 IALLY*OCAMP*repr*INDUSTR*comm*+ growth*INFL*ineq*ENV*URBAN*MOVM* 1.000 0.250 0.125 IALLY*repr*INDUSTR*COMM+ growth*ineq*ENV*URBAN*MOVM*IALLY* 1.000 0.375 0.250 ocamp*repr*INDUSTR*COMM* ➔ EMERGENCE OF PROTEST Solution consistency 1.000 Solution coverage 0.625

Notes: *denotes logical AND, while + denotes logical OR. Uppercase letters denote the presence of a condition, while lowercase letters denote the absence of a condition.

As expected, none of the conditions on its own is sufficient for the emergence of protest after environmental disasters. The conditions are only sufficient for the outcome in combinations. For example, in the absence of environmental culture and state repression, poor economic conditions are linked to protest only in combination with the modernization variables (except communication), the resource variables (except the carrying capacity), and, curiously, low inequality. In a different pathway, poor economic conditions can be linked to protest only in the absence of state repression, and in combination with environmental culture, the resource variables (except the carrying capacity), all modernization variables, and low inequality. The last configuration suggests that a lack of economic growth is sufficient for protest emergence only in combination with environmental culture, resources, favourable political opportunity structure and modernization as well as low inequality. This analysis cannot answer why these combinations (and not others) are sufficient for protest emergence in the aftermath of environmental disasters. A within-case analysis is needed for such explanation. The inclusion of low inequality in all of these pathways goes against theoretical expectations that high levels of inequality and the grievance they generate are linked to protest emergence. On its own, inequality – whether high or low – is not sufficient for the outcome. Its interaction with other variables, however, should be examined in further research.

73 The conditions ‘post-Cold War’ and ‘trade’ were present in all rows and were also removed due to lack of variation.

I have constructed truth tables for the three different protest sizes, as well (see Appendix 2, Tables 35 to 37). For the outcome ‘small protest’, the consistency score in all rows is below 1.0, meaning none of the conditions or their combinations are sufficient for the outcome. The truth tables for the outcomes ‘medium protest’ and ‘large protest’ are more telling. However, the conservative solutions have very low coverage – 50% for the ‘medium protest’ and 33% for the ‘large protest’ (Tables 38 and 39 in Appendix 2). This suggests that, much like with necessary conditions, none of the conditions or their combinations are sufficient for protests to grow into any particular size. The last part of this analysis is the identification of the conditions – both necessary and sufficient – for the absence of post-disaster protest in rich industrialized democracies. I find that none of the conditions qualify as necessary. While some have high consistency values (see Table 13), none has coverage above 50%, suggesting little empirical relevance. The negation of conditions yields the same conclusion.

Table 13. Analysis of necessary conditions for the outcome ‘absence of protest’.

Condition Consistency Coverage ~ Condition Consistency Coverage GROWTH 0.556 0.714 not GROWTH 0.250 0.286 INFL 0.556 0.556 not INFL 0.500 0.444 INEQ 0.444 0.667 not INEQ 0.250 0.333 POV 0.000 NA not POV 0.000 NA ENV 1.000 0.562 not ENV 0.875 0.438 URBAN 0.889 0.500 not URBAN 1.000 0.500 MOVM 1.000 0.529 not MOVM 1.000 0.471 IALLY 0.889 0.500 not IALLY 1.000 0.500 OCAMP 0.222 0.400 not OCAMP 0.375 0.600 REPR 0.333 1.000 not REPR 0.000 0.000 POSTCW 1.000 0.529 not POSTCW 1.000 0.471 INDUSTR 0.889 0.500 not INDUSTR 1.000 0.500 COMM 0.333 0.333 not COMM 0.750 0.667 TRADE 1.000 0.529 not TRADE 1.000 0.471

To identify sufficient conditions for the absence of protest after environmental disasters, I constructed a truth table, presented in Table 14. Subsequently, I derived a conservative solution of six individual combinations of conditions. The first configuration suggests that mixed economic conditions (lack of growth but low inflation and inequality) are sufficient for absence of post-disaster protest only in combination with environmental culture, the resource variables, lack of state repression, and modernization variables without communication. The second configuration suggests that, in the absence of state repression, poor economic conditions (albeit with low inequality) are sufficient for the absence of protest only in combination with environmental culture, pre-existing movement (with no international allies), and no other ongoing campaigns. The third option presents a combination of economic (but not environmental) grievances as well as the resource and modernization variables, but in presence of state repression. The fourth configuration shows that good economic conditions, environmental culture, all resource variables (albeit with other ongoing campaigns), presence of state repression, and modernization variables (with lack of communication) are together sufficient for absence of protest. In the fifth pathway, more favourable economic conditions (albeit with high inflation) are sufficient for the absence of protest only in combination with environmental culture, state repression, established movement with international ties, as well as lack of other ongoing campaigns, urbanization, industrialization and communication. The last pathway presents a combination of some economic grievances, with all resource variables present, environmental culture, lack of repression, and all modernization variables except for communication. The solution term is in line with theoretical expectations about structural conditions and protest – in environments that are typical of advanced industrialized democracies (i.e., lack of state repression, industrialization, modernization, etc.), the absence of some factors such as movement’s international allies or ability to communicate are sufficient for protests not to emerge. There is no single condition that alone can prevent protest emergence. Again, this analysis cannot explain why these particular combinations prevent protest after environmental disasters. It, however, shows that different causal arguments might be necessary to explain the negation of the outcome. The consistency and coverage scores for these conditions are very similar for both presence and absence of the protest outcome, with the coverage score for the latter being slightly higher (67% versus 63%). However, both coverage scores are rather low,

70 meaning they do not capture a large share of the variance of the cases with presence and absence of post-disaster protest. Therefore, additional conditions might be needed in future analysis.74

74 I have not constructed truth tables for the outcomes of ‘absence of small protest’, ‘absence of medium protest’, and ‘absence of large protest’, because these would have not added much explanatory power to the analysis.

Table 14. Truth table for the outcome ‘absence of protest’.

GROW INFL INEQ ENV URB MOV IALLY OCAMP REPR POSTCW

0 0 0 1 1 1 1 0 0 1 0 1 0 1 1 1 0 0 0 1 0 1 1 1 1 1 1 0 1 1 1 0 0 1 1 1 1 1 1 1 1 1 0 1 0 1 1 0 1 1 1 1 1 1 1 1 1 0 0 1 0 0 1 1 1 1 1 0 0 1 1 0 0 1 1 1 1 0 0 1 1 1 1 1 1 1 1 1 0 1 0 0 0 1 1 1 1 0 0 1 0 1 0 0 1 1 1 1 0 1 0 1 0 1 1 1 1 0 0 1 0 1 0 1 1 1 1 1 0 1

INDUS COMM TRADE OUT N Cons Cases

1 0 1 1 1 1.000 12

1 0 1 1 1 1.000 17

1 1 1 1 1 1.000 7

1 0 1 1 1 1.000 13

0 0 1 1 1 1.000 11

1 0 1 1 1 1.000 9

1 1 1 0 2 0.500 5,6

1 1 1 0 2 0.500 4,10

1 0 1 0 2 0.500 14,15

1 1 1 0 2 0.000 1,3

1 0 1 0 1 0.000 16

1 1 1 0 1 0.000 2

1 1 1 0 1 0.000 8

Note: OUT = outcome value, N = number of cases in configuration, Cons. = consistency/sufficiency inclusion score.

Table 15. Conservative solution for the outcome ‘absence of protest’.

Configurations Con. Raw coverage Unique coverage growth*infl*ineq*ENV*URBAN*MOVM*IALLY* 1.000 0.111 0.111 ocamp*repr*INDUSTR*comm+ growth*INFL*ineq*ENV*URBAN*MOVM*ially* 1.000 0.111 0.111 ocamp*repr*INDUSTR*comm+ growth*INFL*INEQ*ENV*URBAN*MOVM* 1.000 0.111 0.111 IALLY*ocamp*REPR*INDUSTR*COMM+ GROWTH*infl*ineq*ENV*URBAN*MOVM* 1.000 0.111 0.111 IALLY*OCAMP*REPR*INDUSTR*comm + GROWTH*INFL*ineq*ENV*urban*MOVM*IALL 1.000 0.111 0.111 Y*ocamp*REPR*industr*comm+ GROWTH*INFL*INEQ*ENV*URBAN*MOVM* 1.000 0.111 0.111 IALLY*ocamp*repr*INDUSTR*comm ➔ ABSENCE OF PROTEST Solution consistency 1.000 Solution coverage 0.667

Notes: *denotes logical AND, while + denotes logical OR. Uppercase letters denote the presence of a condition, while lowercase letters denote the absence of a condition.

Conclusion

The aim of this chapter was to examine whether there are necessary or sufficient structural conditions for the emergence of protest in the aftermath of environmental disasters. The qualitative comparative analysis conducted on 38 cases of disasters yields several findings. First, there seems to be only one necessary condition for the emergence of post-disaster protest in rich industrialized democracies: absence of state repression. Contrary to some literature, other factors such as established movements and underlying grievances do not seem to be necessary conditions for protest. A number of combinations of individual conditions appear to be sufficient for protest emergence and protest absence. This supports the theoretical expectations as well as the prevailing arguments in the social movements literature that no single factor can explain the development of protest. Instead, relevant factors interact in complex ways, including some that are unexpected. Specifically, while high inequality was theorized to be linked to protest emergence, it is the cases with low inequality (in combination with other

73 factors) that are likely to lead to post-disaster protests. The finding suggests that income inequality and its interaction with other structural factors should be explored further. Overall, to understand why and how the proposed combinations of conditions are sufficient for the outcomes requires within-case analyses. For example, process tracing can be applied on typical cases (i.e., cases that are members of both the outcome and the solution) for presence of outcome, and deviant cases for the absence of outcome. Deviant cases can be selected with respect to the statements of necessity and sufficiency or consistency and coverage (Schneider and Wagemann 2012, p. 282). Furthermore, to verify the generalizability of the findings, the original dataset may be expanded to include non-Western democracies or even instances of industrial disasters in non-democratic states. As noted earlier, this analysis reveals that structural conditions are only useful to evaluate the emergence of protest – they do not determine protest size. The protest mobilization process, outlined in Chapter 2, suggests it is framing that ultimately transfers people from the pool of potential protesters to active participants. The following two chapters are therefore focused on framing and its effects on post-disaster protest.

Chapter 5. Disaster Language

“I really do inhabit a system in which words are capable of shaking the entire structure of government, where words can prove mightier than ten military divisions.” Václav Havel (1989)

Disasters have symbolic power in public discourse (Pelling and Dill 2008). From an anthropological perspective, symbolism serves as “a compass of orientation on how to think about calamity and gives an orbit of persuasion on how to cope with and survive it” (Hoffman 2002, p. 114). We invent metaphors and stories to make sense of the world and give life some predictability – an activity that is not necessarily strategic (Oliver-Smith 2002; Oliver-Smith and Hoffman 2002). In politics, however, disaster symbolism often serves strategic objectives. Understanding the language of political actors is central to the study of political effects of environmental disasters. How and for what purposes do these actors frame industrial disasters? What specific dimensions of disasters do they tend to emphasize when pursuing their interests and shaping public opinion? This chapter examines disaster framing as a strategic activity by several political actors, and primarily the governments, corporations, and activists. A principal objective is to better understand the use of specific language and various narratives in the immediate aftermath of industrial environmental disasters. While the use of such language has, to some extent, been explored in several studies, the link between disaster language and protest has not yet been examined. To this end, this chapter presents a text analysis of news media coverage of three major industrial environmental disasters linked to varying degrees of post- disaster protest: the Mount Polley mine leak, the Deepwater Horizon oil spill, and the Fukushima nuclear disaster. The findings reveal both expected and surprising patterns in the post-disaster framing dynamics. Specifically, contrary to theoretical expectations, emotional and environmental frames may play a small (if any) role in post-disaster protest mobilization. The analysis lends some support to the mobilizing potential of relatability (i.e., the idea that the closer and more familiar the event, the more it may matter to us) and prompts further investigation of the role of uncertainty (i.e., under what conditions does uncertainty dampen or encourage post-disaster protest?). Both have been largely neglected as protest mobilizing factors in the prevailing literature. The following four sections explore the theoretical underpinnings of framing, disaster symbolism, and post-disaster uncertainty as an inherent characteristic of disasters. The first two

75 parts include an overview of the framing research, including the definition of frames and framing as well as a discussion of the main framing actors, their interests, and types of frames and tone they are likely to use after environmental disasters. The third part presents a text analysis, including the design, cases, and discussion of findings. The last section consists of a summary and conclusion. This analysis is descriptive – it is about the use of frames in text, not about their effects. I evaluate the effects of frames on target audience in an experiment discussed in Chapter 6.

1. Frames and the Framing Research

Ever since Tversky and Kahneman’s (1979; 1981) demonstration of framing effects that challenged the prevailing concept of rationality, framing has been of steady interest to scholars. It has been increasingly studied across disciplines and subfields including social movements (Tarrow 1994; Benford and Snow 2000; Tarrow 2001), foreign policy and international relations (Levy 1997; Mintz and Redd 2003; Boettcher 2004; Mercer 2005; Perla 2011), as well as political behavior and media effects (Neuman, Just, and Crigler 1992; Chong 1996; Price and Tewksbury 1997; Sniderman and Theriault 2004; Brewer and Gross 2005; Lim and Seo 2009). While there are many different definitions and classifications of frames,75 this section’s goal is to briefly outline some broader conceptual distinctions necessary for an understanding of what frames are, and how they may affect the public’s preferences to mobilize for protest. Framing affects behavior not because of what is being communicated but how a piece of information is being presented in public discourse (Druckman 2001; Scheufele and Iyengar 2014). Frames are cognitive shortcuts, and framing is a process through which people make sense of the information they receive based on some pre-existing ‘schemas’ (Benford and Snow 2000; Kwan 2009; Scheufele and Iyengar 2014).76 In psychology, frames are generally

75 See, for example, Benford and Snow 2000; Mintz and Redd 2003; Keren 2011. 76 Some scholars call for more specific distinctions between framing and other modes of public influence (Scheufele and Iyengar 2014), while others believe the distinction is unnecessary as long as the effects are the same (Druckman 2011). Framing is generally contrasted with two main modes of public opinion formation: agenda-setting and priming. The former assumes that effective messages need to resonate with the underlying ‘cognitive schemas’ that the target audience holds (Price and Tewksbury 1997; Scheufele 2004; Scheufele and Tewksbury 2007). The importance of the message is transferred from the mass media to target audiences (McCombs and Shaw 1972; Edelstein 1993). Priming is based on a similar logic; it refers to the process that begins after an issue is presented as highly important to target audiences where the main assumption is that such presentation activates previously learned cognitive structures and therefore influences preference formation (Collins and Loftus 1975). 76 understood in a restrictive way as “informationally equivalent labels” (Keren 2011, p. 5; Scheufele and Iyengar 2014). In this sense, framing is about different ways of presenting the same piece of information – often in negative or positive light (i.e., equivalence framing) (Levin, Schneider, and Gaeth 1998; Druckman 2011; Scheufele 2014).77 An alternative type of framing is emphasis framing, which is about conveying different perspectives on an issue or event – it is about how an issue should be understood (Druckman 2001; Scheufele and Iyengar 2014). In sociology, a frame can be viewed in more encompassing terms as a ‘packaging’ of information (Gitlin 1980), a message or a storyline that provides meaning to events (Gamson and Modigliani 1987; see also Keren 2011, p. 4). Frames can be further studied from two perspectives – the position of the framer and the view of the target audience. For example, Druckman (2011) divides frames into two general categories: ‘frames in communication’ (i.e., what is being said – words, phrases, images or presentation styles) and ‘frames in thought’ (i.e., what one thinks – one’s understanding of an event or issue). Social movements scholars often study frames from the perspective of the framing actor – in terms of what frames are meant to achieve, not necessarily how they are perceived (Gamson and Meyer 1996, p. 283; Zald 1996, p. 262; Benford and Snow 2000). Other scholars have also been more interested in framing actors and frames based on their intended rather than actual effect – for example, frames are meant to define problems, diagnose causes, make moral judgments, and suggest solutions (Entman 1993; Norris 1995; Entman 2003), but whether and under what conditions they do is less clear. Since this research is primarily concerned with public reactions to environmental disasters, it focuses on both the framing actors (including their interests and framing strategies) and the target audience (i.e., potential protesters). This chapter and the following sections discuss the former. The target audience and effects of specific frames on public willingness to participate in protest is explored in Chapter 6.

2. Disaster Framing and Post-Disaster Uncertainty

Political actors frequently use metaphors, targeted phrases, carefully chosen words, or pictures to use disasters for “political positioning” (Pelling and Dill 2009; see also Pelling and

77 Tversky and Kahneman’s (1981) ‘Asian disease problem’ is a famous example of equivalence framing. In an experiment, participants were presented with choices framed as the number of lives saved, given certain population (200 lives out of 600 people) vs. the number of lives lost (400 lost out of 600).

Dill 2006). Language and symbols help these actors construct beliefs about the significance of events, issues, policies, and leaders (Edelman 1985). Through symbols used to describe and/or explain the disaster, political actors give the event “context, content, emotion, and meaning” and thus influence shared behavior (Hoffman 2002, pp. 114–115). Utilization and manipulation of symbols turn disaster political at both international and domestic levels. For example, governments use specific ways to talk about disasters (and especially very destructive events) to achieve different objectives: the language of ‘tragedy’ and ‘devastation’ to attract international assistance (Nebehay 2011), the imagery of government competence to boost its popularity (Chen 2009), the rhetoric of ‘building back better’ to initiate domestic social reforms (Schencking 2008; Fan 2013), or specific portrayals of disaster relief recipients for donors’ own purposes (Wilder 2010). At a domestic level, some major disasters become symbols that open political opportunities or fuel social mobilization. For example, the Chernobyl nuclear disaster provided opportunities for dissent against governments in the Soviet republics (Petryna 1995; Dawson 1996; Larabee 2000, p. 46). The Exxon Valdez oil spill was also heavily symbolized by different groups. For example, environmentalists made it into a symbol of the nation’s destructive obsession with oil (Larabee 2000, p. 83). The spill is one of the most well-known cases of framing and counter-framing by different political actors struggling to delegitimize each other’s claims in order to sway public opinion to their side (Daley and O'Neill 1991; Smith 1993; Larabee 2000; Lendon and Martin 2007). The possibility of such framing and counter-framing is, to an extent, created by the uncertainty that disasters generate. Uncertainty characterizes the immediate aftermath of any disaster. What caused the event? What are the consequences? How will the damage be addressed? Will this happen again? Before the framing dynamics can unfold, political actors must respond to the uncertainty that the event causes. The effectiveness and ways in which they do will likely affect their subsequent activities. Political actors may also shape and use post-disaster uncertainty to advance their interests. Uncertainty is therefore a crucial element of post-disaster political dynamics (see, for example, De Marchi et al. 1996). For example, after the Chernobyl disaster, scientific and institutional uncertainty significantly shaped the nuclear policies and learning in different European countries (Liberatore 1999). Aside from a few studies of uncertainty in environmental discourse (Morton et al. 2011; Bailey, Giangola, and Boykoff 2014; Blair et al. 2016) and political behaviour (Alvarez and Franklin 1994; Haas and Cunningham 2014), political effects of

78 uncertainty have not yet received much attention from scholars. This chapter adds uncertainty as a factor in post-disaster framing. Industrial disasters generate uncertainty that can be broadly categorized into two types: inherent and contextual. Inherent uncertainty is rooted in an incomplete or imprecise information surrounding the event, and it is linked to human inability to assess a situation, make predictions, and take action (Marris 2005). Inherent uncertainty therefore stems from external sources beyond the control of political actors. It can be situational (“involves a poor match between the decisions that must be taken and the information at hand”) or scientific (linked to difficulties with risk assessment and effective forecasting) (De Marchi et al. 1996, p. 97). Two important aspects of inherent post-disaster uncertainty are lack of clarity regarding the disaster cause,78 and uncertainty about the extent of damage, including both immediate and future harm. In contrast, contextual uncertainty results from deliberate actions of political actors. It can be institutional (when government agencies withhold information for bureaucratic reasons) or legal/moral (due to insufficient or delayed release of information because actors fear that they may be subject to legal action) (De Marchi et al. 1996, pp. 97–98). Lack of transparency and various cover-ups79 linked to some nuclear accidents in Japan (including the 1999 Tokaimura and 2011 Fukushima disasters) are examples of such contextual uncertainty (Beech 1999; Aldhous and Iovino 2011). Another type is reinterpretation of the event – actors claiming the event or some aspects of it are not what they seem. Different political actors may “deny that an act occurred, deny knowledge of the act, deny that the action meant what others thought it did, or deny any intention to cause the act” (Martin 2007, p. 4). In the end, the difference between contextual and inherent uncertainty depends on the perspective. For example, if the government has the information on disaster damage but chooses to withhold it from the public, from the government’s perspective the inherent uncertainty is minimal, but from the perspective of the public, the uncertainty can be considerable. In fact, withholding information from the public creates contextual uncertainty. Since this research is concerned primarily with public reactions to disasters, inherent uncertainty may be of less

78 The cause has two dimensions: what happened and why did it happen? What happened is usually obvious – for example, a tailings pond broke, an oil tanker crashed, or a pipeline burst. This type of unclear cause is likely to be rare for acute industrial disasters. On the other hand, why a disaster happened is often unclear in the immediate disaster aftermath, and it takes months or even years to determine. These types of causes include, for example, intentional sabotage, lack of funding to invest into safety measures, or negligence. 79 Different types of cover-up include censorship as well as hiding, destroying or refusing to collect relevant data (Martin 2007). 79 importance unless such uncertainty is universally prevalent. Contextual uncertainty, however, is a result of framing strategies of different political actors. They can increase uncertainty through specific actions such as cover-ups or reinterpretation or directly frame the event as uncertain with carefully chosen words, therefore using uncertainty as a rhetorical strategy. The following sub- sections discuss in more detail who these actors are in the context of environmental disasters as well as their interests and types of frames they are likely to use in the disaster aftermath.

2.1 Framing actors

In politics, framing is used by multiple actors attempting to influence political decisions in specific contexts – through distributing competing information and, consequently, shaping public preferences that matter in electoral outcomes and policy decisions (Zald 1996, p. 262; Erikson, MacKuen, and Stimson 2002; Entman 2003; Druckman 2011; Scheufele 2014). Political actors struggle for power to control information in public discourse and to define reality in accordance with their interests, legitimizing some accounts of the event and privileging some political agendas over others (i.e., the processes of framing and counter-framing80) (Zald 1996; see also Chong and Druckman 2012). After industrial disasters, three political actors are likely to be the predominant producers of frames: activists, governments, and the responsible corporations. While the latter two will always be present and influential in the post-disaster81 social environment, the influence of activists will be conditional upon the strength of the pre- existing anti-industry or environmental movement. Although activists, and social movement organizations in particular, tend to have a range of interests (see, for example, Fisher 1997; Tvedt 1998; Kellow 2000), in the context of this study, only one likely objective is explored: mobilization of the public for protest.82 Industrial disaster scholars tend to group together governments and corporations, assuming they have common objectives (Gephart 1984; Birkland 1998). However, corporate interests and subsequent framing of disasters differ from the government ones. After an industrial disaster, corporations will want to control the disaster damage, re-stabilize their public image,

80 Counter-frames are frames in opposition to original frames. 81 This, of course, applies only on industrial disasters. Corporations will not always be a principal actor in the aftermath of natural disasters. 82 This focus, however, limits the analysis, since lack of protest could simply be explained by activist unwillingness or inability to mobilize the public. Field interviews are needed to establish activist motivations and resources for protest mobilization in specific cases of industrial environmental disasters.

80 and maintain the policy status quo (i.e., prevent stricter regulations) (Birkland 1998; Larabee 2000; Sindermann 2005; Breeze 2012). At a general level, government interests seem no different. At a policy level, however, the government objectives may differ significantly. Because disasters draw public attention to apparent policy failures, one of their consequences is the erosion of public trust in the government (Birkland 1998; Adeola 2011). Therefore, government frames will likely be constructed to serve government’s primary interests: to remain in power, prevent a decline in its legitimacy, and implement policies in line with government’s preferences (Weaver 1986; Levin 2005; Lim and Seo 2008; Hannigan 2012). Government interests are unlikely to be homogeneous. Inter-agency differences, divisions between legislative and executive interests, and tensions due to federalism, for example, may result in different framing efforts among different government actors. In a pursuit of their interest, the three framing actors use different types of frames to influence their target audience. In this process, communication moves from the political elites to the media to the public, with the news media serving as a principal conduit for framing competition (Klar, Robison, and Druckman 2013; Scheufele and Iyengar 2014). Newsmakers, having their own motivations and biases, construct reality (and therefore disasters)83 and thus influence public perceptions and political behavior (Benthall 1995; Boykoff and Boykoff 2004; Carvalho 2007; Shanahan, McBeth, and Hathaway 2011). However, unlike government and corporate frames that are generally meant to facilitate or constrain certain outcomes, media frames are largely based on newsroom routines (Levin 2005; Zald 1996, p. 270). Industry spokespeople and other groups decide which statements will be released to the media – they control the information flow (Entman 2003). Therefore, this research treats the news media primarily as a means of distributing strategic messages from political elites.84 The following sub- section examines the themes and the tone of coverage likely to appear in the news media after a disaster occurs.

83 The role of news media during crises and disasters has been widely studied. See, for example, Molotch and Lester 1975; Farrell and Goodnight 1981; Perez 2003; Anderson and Marhadour 2007. 84 However, since journalists choose and structure stories to present them to the public in specific ways, they are included in this analysis as a category of framing actors. 81

2.2 Types of post-disaster frames

The most typically used frames in public discourses can be organized into five broad thematic categories: responsibility, human interest, morality, economic consequences, and conflict (Semetko and Valkenburg 2000; see also Koenig 2004; de Vreese et al. 2010). These can, however, be expanded to better account for the post-disaster framing dynamics. Two additional types of frames are likely to occur in the news media after industrial environmental disasters: environmental frames, and industry frames linked to the public discourse concerning the energy industry at fault. First, responsibility frames assign responsibility for the problem’s cause to an individual, group or government (Neal 1984; Semetko and Valkenburg 2000). These types of frames are a common feature of the post-disaster dynamics where assignment of blame is a principal task performed by political actors (Bucher 1957; Drabeck and Quarantelli 1967; Gephart 1984; Neal 1984; Gephart 1993; Javeline 2003; Waugh 2006; Catino 2008; Pantti and Wahl-Jorgensen 2011). For example, diagnostic framing (i.e., defining the problem and assigning blame) is one of the core framing tasks of social movement organizations (Snow and Benford 1988; Benford and Snow 2000). Corporations and governments are more likely to employ “the vernacular of damage control” (Olson 2008, p. 163): excuses and justifications (see also McGraw 1991). The former is about denying – partially or fully – one’s responsibility. The purpose of the latter is to create “an alternate political reality” in terms of reframing the undesirable issue in a more favorable light (McGraw 1991, p. 1137). Second, the human interest frames are meant to trigger an emotional response from the target audience; they often put ‘human face’ on the problem at hand, dramatizing it in order to make the problem more personal (Semetko and Valkenburg 2000). Emotions are at the core of social movements and have a crucial role in activist framing (see, for example, Wood 2003). Emotions are believed to affect the ways in which people process frames; they are “an implicit perceptual lens for interpreting situations” (Druckman 2008; see also Lerner and Keltner 2001; Lerner et al. 2003). Furthermore, both psychological and framing research suggests that specific emotions either encourage or discourage individuals from action (Johnson and Tversky 1983; Marcus et al. 2000; Berkowitz and Harmon-Jones 2004a, 2004b). Emotionally charged frames are therefore likely to be present in the post-disaster framing dynamics.

Morality frames are the third category of post-disaster frames. Their purpose is to place the problem in the religious context or make some moral or ethical prescriptions (Semetko and Valkenburg 2000). Framing actors may refer to the notions of ethics, the right or wrong, and various social norms. In a post-disaster environment, actors may attempt to blame the event on God or some natural processes or claim that the event was impossible to prepare for. The narratives about the damage are therefore placed outside of human control, deflecting blame from the governmental and corporate actors (Button 2002). Fourth, economic frames emphasize the economic dimensions of the problem, often in terms of economic impacts on individuals, groups or country (Semetko and Valkenburg 2000). These frames present material gains or losses and various trade-offs of the issue at hand (Karlberg 1997). After an industrial disaster, economic frames may focus on the economic losses (or benefits) of the industrial activity that caused the disaster or emphasize the economic impacts of the disaster itself. Fifth, conflict frames reflect varying degrees of conflict between individuals, groups or institutions (Semetko and Valkenburg 2000). These frames’ main characteristics are dichotomy (i.e., the problem is framed from a perspective of two distinct, mutually exclusive, stereotypical camps) and extremism (i.e., dramatization of conflict through emphasis on extreme statements and actions; it includes insults, accusations, or angry expressions) (Karlberg 1997). In the post- disaster dynamics, such frames may also include accusations of government and/or corporate cover-ups or various relevant wrong doings. The sixth type of frames likely to appear after industrial disasters are environmental frames. As per the prevailing literature, environmental frames tend to convey the issues of environmental justice (for example, Kurz 2003), disaster preparedness and climate change (Trumbo 1996; Morton et al. 2011; Bertolotti and Catellani 2014; Vinnell, McLure, and Milfont 2017) or post-disaster narratives (Molotch and Lester 1975; Luke 1987; Smith 1993). In the post- disaster context, environmental frames are likely to emphasize harmful environmental impacts of the disaster, including the immediate damage and possible ongoing or future harm. In general, the damage could be framed in two ways: as ‘natural’, for example through comparisons of the industrial disaster to natural processes (Karlberg 1997), or as human caused. The former way of framing ‘naturalizes’ the disaster and removes it from human responsibility, making it seem inevitable (see Perrow 1984). The latter links the disaster damage to the human factor – either to

83 the specific circumstances of the event or to a broader trend (e.g., industry focus as a whole or climate change narratives). Lastly, some aspects of the national energy policies are also likely to be reflected in post- disaster framing. These can be viewed through the lens of long-time public discourses with three sides with opposing narratives: pro-industry, anti-industry, and neutral/indifferent. Table 16 presents these industry-specific frames as well as the remaining six themes along with a series of questions used to evaluate the frames’ presence or absence in the news coverage. The questions have been adapted from Semetko and Valkenburg (2000) and Giannopoulos (2013).

Table 16. Post-disaster frames in news coverage.

1. Responsibility frames Assignment/Acceptance of blame Q1a. Does the framing actor suggest that some level of government, corporation, individual or group is responsible for the disaster? Q1b. Does the framing actor suggest that some other thing (without referring to a specific actor – e.g., a faulty design, mechanical problems) is responsible for the disaster? Q1c. Does the framing actor suggest that some level of government has (or has had) the ability to alleviate the problem? Q1d. Does the framing actor suggest that the pre-existing practices (governmental or corporate) are not working? Q1e. Does the responsible actor accept responsibility for the disaster?

Denial of blame Q1f. Does the framing actor employ excuses (i.e., deny responsibility)? Q1g. Does the framing actor employ justifications (i.e., place the issue in a more favourable light)? Q1h. Does the framing actor suggest the disaster was ‘an act of God’, a natural occurrence, or otherwise was impossible to predict?

2. Human interest frames Q2a. Does the framing actor emphasize how individuals and groups are affected by the disaster? Q2b. Does the framing actor discuss the personal or private lives of the impacted individuals? Q2c. Does the framing actor employ adjectives, metaphors or anecdotes that generate feelings of fear, anger, empathy or sympathy? Q2d. Does the framing actor refer to their own personal experience (e.g., their own families affected or potentially affected in the future) or the humankind?

3. Morality frames Q3a. Does the framing actor imply (un)ethical or (im)moral actions on the part of themselves or others? Q3b. Does the framing actor offer specific social prescriptions about how to behave either in the short term or long term? For example, does the framing actor urge others to help the affected people out of a sense of moral duty?

Q3c. Does the framing actor refer to morality, God, or other religious tenets?

4. Economic frames Q4a. Does the framing actor refer to the economic costs of the disaster (in monetary or non- monetary terms, actual or potential) to individuals, groups, regions or the country, in the immediate aftermath or in the future? (This includes the costs to the responsible company such as the cost of clean-up, various costs to taxpayers, or the company pledging funds for research linked to the disaster.) Q4b. Does the framing actor refer to economic consequences that are explicitly linked to the environmental damage from the disaster (e.g., the cost of lost subsistence)? Q4c. Does the framing actor mention economic consequences of pursuing or not pursuing a course of action such as specific policies related to the disaster? Q4d. Does the framing actor employ adjectives, metaphors or anecdotes that generate feelings of fear or anger linked to the economic consequences of the disaster?

5. Conflict frames Q5a. Does the story reflect disagreement between individuals, groups, different levels or parts of government or other actors? Q5b. Does the framing actor criticize or accuse another? Q5c. Does the framing actor refer to others as extremist or assign other dramatic labels? Q5d. Does the framing actor refer to clear dichotomies such as jobs vs. conservation or economy vs. the environment? (These need not be related to the economy.)

6. Environmental frames Q6a. Does the framing actor mention the current, ongoing or future environmental damage (and/or its consequences) caused by the disaster? Q6b. Does the framing actor link the disaster damage to broader themes such as the deteriorating state of the environment or climate change? Q6c. Does the framing actor employ adjectives, metaphors or anecdotes that generate feelings of fear or anger linked to the environment and loss of environmental values? (This could also include referring to extreme or never before seen damage, or using words such as emergency, catastrophe, etc.) Q6d. Does the framing actor suggest the environmental damage, although present, is small, minimal, insignificant, contained or not as large as it seems?

7. Industry-specific frames A. Pro-industry Q7a. Does the framing actor refer to economic or environmental benefits of the industry or the industry’s importance as a whole (e.g., the oil & gas industry, the mining industry, etc.)? Q7b. Does the framing actor refer to stakeholder support for the industry? Q7c. Does the framing actor refer to the reliability, safety and/or security of the specific energy?

B. Anti-industry Q7d. Does the framing actor suggest economic, environmental, or health/safety risks of the industry independent of the disaster itself? Q7e. Does the framing actor refer to a need for alternative energy or calls for the energy phase- out?

Q7f. Does the framing actor refer to previous (or potential future) accidents or disasters (similar or not) caused by that particular industry? Q7g. Does the framing actor employ adjectives, metaphors or anecdotes that generate feelings of fear or anger linked to the particular industry?

C. Neutral/indifferent Q7h. Does the framing actor suggest a balance when it comes to the advantages and disadvantages of that particular energy production? Q7i. Does the framing actor suggest an inevitability of that particular energy production? Q7j. Does the framing actor take an undecided or no position on the industry energy? Q7k. Does the framing actor refer to a trade-off between the type of energy and other issues?

In addition to specific themes, this research also evaluates the tone of frames used by the framing actors after environmental disasters, because tone is likely to stir up or dampen specific emotions linked to willingness to protest (see, for example, Bennett 1980). Tone is generally evaluated as positive or negative (Grimmer and Stewart 2013), but some studies have developed more nuanced tone measures in newspaper coverage (Burgoon 1978; Pfau et al., 2004; Brunken 2006; Haigh et al. 2006). In this study, the descriptors of tone were adopted from Haigh et al. (2006), Brunken (2006) and Giannakopoulos (2013), and adjusted to better reflect the post- disaster environment (see Table 17). In news coverage, tone is captured in statements by framing actors as well as in descriptions by the media.

Table 17. Tone of post-disaster frames.

a) Absent (0) – Unsuccessful (1) – Successful (2) – Neutral (3) – Unclear (4) b) Absent (0) – Unprepared (1) – Prepared (2) – Neutral (3) – Unclear (4) c) Absent (0) – Unreliable (1) – Reliable (2) – Neutral (3) – Unclear (4) d) Absent (0) – Obscure (1) – Informative (2) – Neutral (3) – Unclear (4) e) Absent (0) – Uncertain (1) – Certain (2) – Neutral (3) – Unclear (4) f) Absent (0) – Relatable (1) – Unrelatable (2) – Neutral (3) – Unclear (4)

The successful/unsuccessful tone refers to the government officials’ and/or company’s handling of the disaster in its aftermath. A successful tone, for example, is reflected in references to a speedy response with proper cleanup procedures. The unprepared/prepared tone conveys the preparedness for either that particular disaster or similar disasters or disasters as a whole. References to weak regulations, for example, suggest lack of preparedness. The unreliable/reliable tone refers to the government’s or corporate actor’s degree of trustworthiness and dependability. The obscure/informative tone is linked to the framing actors either obscuring or providing information on the disaster, including the cause, responsible actors, and the extent

86 of damage. This tone is specifically about the framing actor’s willingness (or lack of it) to provide information about the disaster. The uncertain/certain descriptor goes a step further – it captures the use of uncertainty as a rhetorical strategy using, for example, specific words (e.g., potentially, probably) that denote uncertainty (see, for example, Dunwoody 1999; Bailey et al. 2014; Windsor, Dowell, and Grasser 2014). Lastly, the relatable/unrelatable descriptor captures framing attempts to relate (or not) the disaster to the target audience.85 Details on the specific types of tone are available in the codebook (Appendix 3). The preliminary hypotheses presented in the next sub-section link these different types of tone and frames to framing actors and protest outcomes.

2.3 Expectations

To my best knowledge, there is no established literature on the topic of disaster language and protest that could be used to guide my hypotheses. The following analysis is, therefore, more inductive, and the inferences are correlational, not causal. The frames explored here are frames in communication only (frames in thought are utilized in Chapter 6). Furthermore, I assume that due to the multidimensional nature of industrial environmental disasters, emphasis frames will be more common than equivalence frames. Equivalence frames are used in experimental research and are unlikely to have a strong presence in the news coverage. Therefore, this analysis focuses on emphasis frames and specifically as they relate to the seven themes discussed in the previous section. All these themes are likely to appear in the news coverage to some extent, but different aspects of these frames would be emphasized by different framing actors. The responsibility frames could be divided into two groups: assignment of blame and denial of blame (through excuses and/or justifications). Activists are likely to assign blame for the event, while governments and corporations are expected to engage in some type of denial to deflect blame. Different levels of government, however, are likely to differ in their use of responsibility frames. Unlike incumbents, opposition parties generally lack incentives to blame the corporation, and are instead expected to challenge the party in power. Local governments – if

85 This is closely linked to the concept of the so-called “communities of interest” (Birkland 1998, pp. 54–55). These are communities that may be thousands of kilometers away from the disaster, but whose members identify with the harm that the disaster caused, believe that a similar disaster may occur in their own community sometimes in the future, and often fear such a possibility.

87 directly affected by the disaster – are likely to blame higher levels of government and/or those in charge of regulating the industry. Since human interest frames tend to trigger an emotional response, they are more likely to be used by activists and journalists than by governments and corporations. Morality frames, too, are more likely to be used by activists, since implying immoral or unethical behavior is an effective way to mobilize the public for protest – these frames not only attribute responsibility for a disaster, but also nefarious motives to corporate or government actors. Governments and corporations may attempt to frame the disaster as an event out of their control (i.e., blame it on God). They are also likely to minimize the cost of disaster impacts. Activists, however, are likely to focus on the costs of disasters – economic or otherwise, since emphasizing such costs may to appeal to non-environmentalists, as well. With respect to industry frames, activists are expected to take the anti-industry stance, while corporations are linked to pro-industry narratives. Governments could take any side, including the neutral position, with the choice depending on domestic politics and specific government agencies involved. All relevant framing actors are likely to use conflict and environmental frames – where they would differ is the tone. While activists are likely to adopt negative tone to mobilize the public, corporate and government frames are expected to be more positive. The following section presents the design, procedures, cases, and findings from a content analysis used to evaluate these hypotheses.

3. Content Analysis

Because of its suitability for analyzing large bodies of text (see, for example, Deacon et al. 2007), content analysis is a fitting technique for meeting the objectives of this chapter. Content analysis is “a family of research techniques for making systematic, credible, or valid and replicable inferences from texts and other forms of communication” (Drisko and Maschi 2011). It is usually used in exploratory or descriptive research design but has found application in hypothesis testing as well (Krippendorff 2013).86 The primary purpose of content analysis in this chapter is to assess the types of frames that different political actors produce after industrial disasters. To this end, three cases have been selected for analysis: the Mount Polley mine leak,

86 For other applications of descriptive content analysis, see Riffe, Lacy and Fico (1998, pp. 14–20).

88 the Deepwater Horizon oil spill, and the Fukushima nuclear disaster. These cases, along with the rationale for their selection, are discussed below.

3.1 Case selection and cases

These cases were selected based on a number of reasons. First, pragmatically, the three disaster events occurred within a few years of one another, which allows for controlling for some underlying structural conditions such as broader social, political, economic, and technological environment. Second, since the disasters generated substantial media coverage, they are data-rich cases, and their examination is likely to lead to more meaningful findings. Third, these cases are also the largest disasters in their respective categories to date; yet, they were followed by different scales of protest – from small and localized after the Mount Polley leak to medium-size protests after the Deepwater Horizon spill to mass protests in Germany after the Fukushima disaster. A brief description of each case follows. The Mount Polley disaster was a mining leak that occurred on August 4, 2014 in the Cariboo region of British Columbia. Between fifteen and twenty-four million cubic meters of mine waste (mostly arsenic and mercury) polluted nearby lakes and rivers, making the Mount Polley event the largest mining disaster in Canadian history and one of the worst mining environmental disasters in the world (Lee 2014; Meissner 2016). The open pit copper and gold mine is owned by a Canadian company, Imperial Metals, who has claimed responsibility and later has been found at fault by an independent, government-ordered expert panel. Poor practices (specifically, an inadequately designed dam for the tailings pond) were determined as the cause of the disaster over five months after the event (Hunter and Hume 2015; Linnitt 2015; Meissner 2016). No casualties were reported, but the environmental damage was large, primarily due to the spilled mine waste clogging the salmon-bearing habitat. Long-term impacts are uncertain – the type and extent of harmful effects on fish, plants and insects due to exposure to metals are unclear (Hume 2015). Some nonviolent protests followed the Mount Polley disaster. These occurred in Vancouver about a week after the event. The protests were mostly aimed at the mining company and were small in size (Richmond 2014). In addition, the Mount Polley disaster appears to have had some transnational effects, giving “ammunition” to Alaskans who had been concerned about

89 potential pollution from a proposed copper-gold mine (the KSM project) in northwestern British Columbia (Keller 2014; Bennett 2016). The Deepwater Horizon disaster was a massive oil spill due to an explosion of a mobile offshore drilling rig about 66 km off the coast of Louisiana in the Gulf of Mexico. The explosion occurred on April 20, 2010, damaging a wellhead some 1,500 meters below surface. The leak continued until July 15, 2010 when the wellcap was replaced. By then between 500 and 600 thousand tonnes of oil leaked into the Gulf, making the Deepwater Horizon disaster the largest accidental oil spill in history. The explosion resulted in death of 11 platform workers and injuries of several others (Hoffman and Jennings 2010). Environmental damage was immense – almost 1,300 km of coastal habitat, including wetlands and beaches, were oiled, severely affecting seabirds, marine mammals, fish and corals (Freudenburg and Gramling 2011; Bishop 2014; NOAA). The company that had contracted to use the drilling rig, the British Petroleum (BP), assumed responsibility for the disaster (Kerr et al. 2010). Nonviolent protests – demonstrations, petitions and boycotts – against BP erupted across the United States and Britain (Jonsson 2010; Martin 2011). Despite strong established environmental movement and heavy medialization, the protests that erupted after the Deepwater Horizon spill were only middling in size (Klaus 2010; Wheaton 2010; Schmidt 2014). In the end, BP agreed to pay $18.7 billion in settlement, the largest environmental fine in history (Rushe 2015). The Fukushima nuclear disaster was a meltdown of the Fukushima Daiichi nuclear power plant operated by Tokyo Electric Power Company (TEPCO). The disaster followed an earthquake and tsunami that occurred on March 11, 2011 about 100 km off the northeast coast of Japan. The failed reactors were releasing radioactive material for months before plant operators managed to stabilize them in December 2011 (BBC 2011). No deaths were recorded due to radiation exposure, but twenty people died due to explosions and during evacuation (Elliott 2013, p.8). Environmental impacts of the disaster were devastating – radioactive material contaminated the mainland and flowed into Japan’s coastal estuary systems and the Pacific Ocean (Greenpeace 2016). Uncertainty surrounded the amount of radiation released, the impacts of radiation on the environment and nearby residents, and the future effects (Brumfiel 2012; Tanaka 2012). Although the Fukushima disaster impacted countries across the world, public responses in two of them were in an especially stark contrast: post-disaster protest mobilization in Germany versus that in Japan. In Germany, tens of thousands of people took to streets days after the

90 disaster, calling for a national nuclear phase-out. On March 12, 2011, one day following the nuclear meltdown, 60,000 protesters formed a human chain stretching for 45 kilometers between Stuttgart and the Neckarwestheim nuclear power plant (Hasegawa 2014, p. 291). On March 14, 110,000 people protested across 450 municipalities, and on March 26, 260,000 people across Germany gathered in the largest antinuclear demonstration in German history (with over 100,000 protesters in Berlin) (Hasegawa 2014, p. 291). In Japan, the protests began a month after the disaster and were constrained to Tokyo – 15,000 people gathered on April 10, 20,000 on June 11, and 64,000 on September 19 (Elliott 2013, p. 18; Hasegawa 2014, p. 292). These protests, however, were short-lived, amateur, and largely failed to gain political traction (The Economist 2014). Although the content analysis described in the following two sub-sections does not establish any causal links between these three disasters and the size of protests that followed, it sheds light on the political use of ‘disaster language’, thus contributing to a better understanding of the conditions under which such protests are likely or unlikely to develop.

3.2 Data sources and coding

In general, content analysis involves drawing a representative sample of the content of interest, developing coding rules (and training coders) to measure differences in content, and measuring the reliability of coders (i.e., intercoder reliability) in applying the coding rules (Riffe, Lacy, and Fico 1998, p. 3; Grimmer and Stewart 2013). Using LexisNexis Academic, I have collected all available newspaper articles produced after the disaster, using the following keywords: Mount Polley; Deepwater Horizon OR (BP AND oil spill); Fukushima. The search dates spanned from the day of the disaster occurrence to its anniversary date a year later. The time period is somewhat arbitrary but is likely to cover all significant protest events that occurred in the immediate disaster aftermath and before the recovery period (which has a different social dynamic than the immediate aftermath). At the same time, this delineation excludes anniversary protests and broader anti-industry protests that may have less to do with the disaster itself. Rather than selecting specific newspapers for analysis, at first, I included all relevant articles appearing in any newspaper during the specified time period. The selected news coverage was case specific – Canadian for Mount Polley, US for Deepwater Horizon, and German for Fukushima. Since I am interested in political actors’ attempts to influence public opinion, I did not intentionally exclude commentaries and op-eds. Next, I narrowed the article

91 population to include only newspapers with the highest readership, accounting for regional coverage to more accurately reflect countries’ media agendas. I have kept articles in any of the nine major Canadian newspapers (see Soroka 2002):87 the Globe and Mail, Toronto Star, Montreal Gazette, Halifax Chronicle, Calgary Herald, Vancouver Sun, Winnipeg Free Press, and La Presse. I have also included articles in major newspapers close to the disaster zone such as the Prince George Citizen as well as major provincial newspapers such as the Province and the Times Colonist. Similarly, I have kept articles appearing in major national and regional US newspapers: USA Today, the New York Times, the Wall Street Journal, Los Angeles Times, New York Post, Chicago Tribune, the Washington Post, Newsday, Daily News, am New York, San Francisco Chronicle, and St. Louis Post-Dispatch. I have also included the largest newspapers (by circulation) in the states affected by the disaster: the Dallas Morning News, Houston Chronicle, the Birmingham News, Baton Rouge Advocate, Tampa Bay Times/St. Petersburg Times, Tampa Tribune, the Clarion-Ledger, and Star-News (Northern Carolina). Since in Germany the local and regional press is more important than national newspapers (see Kleinsteuber and Thomass 2007), I did not apply the same exclusion criteria as for the Canadian and US coverage. From the initial dataset I have only eliminated the Swiss press as well as a couple of small publications (Manager Magazine and Aar-bote). Along with major regional and local newspapers (e.g., Berliner Zeitung, Kölnische Rundschau, Stuttgarter Nachrichten, Spiegel) the dataset contains the main national papers in Germany, including Welt, Frankfurter Rundschau, and die Tageszeitung. After applying the above exclusion criteria, the total population was 1,547 articles, of which 543 were about Mount Polley, 329 about Deepwater Horizon, and 677 about Fukushima. For each case, I drew a random sample of approximately one third of the articles. The total population of sampled articles was 437, of which 200 were about Mount Polley, 111 about Deepwater Horizon, and 226 about Fukushima.88 I have coded articles manually to measure their content. Hand-coding involves categorizing a set of documents by hand – according to the coding rules developed for this purpose. I first created a coding scheme (Appendix 3) and applied it to an initial sample of

87 Soroka (2002) has argued that these newspapers together are representative of Canada’s regions and owners, and are therefore a good indicator of the Canadian newspaper agenda. 88 During the coding process, articles that not contain a substantial section on the disaster of interest were not coded.

92 articles to correct for any ambiguities or overlooked categories. I then revised the codebook and applied it to another set of documents. I have adjusted the codebook in this way three times. The codebook includes basic identification information about each article (title, author, date, source), and instructs the coder to answer the sets of questions in Table 16 as they relate to specific frames. In addition, the codebook instructs the coder to note whether any actors quoted or paraphrased in the article were linked to specific frames and tone. There are nine categories of framing actors: journalist, activist, local government, provincial/state government, federal government, corporation, expert, other, and unclear.89 The news articles were coded by two trained coders. Intercoder reliability test was conducted by randomly selecting 10% of the articles from the sample. Intercoder reliability was 97%, using Holsti’s formula that measures the percentage of agreement (Holsti 1969, p. 140) and 0.83, using Krippendorff’s alpha (Krippendorff 2013). Both suggest a suitable level of intercoder agreement. In addition to hand-coding, I used computer assisted data clustering (i.e., topic modelling) to see whether the frames contained in my full population of articles are consistent with the hand-coded sample, or whether there are any categories I may have missed. Topic modelling involves the use of algorithms that analyze the words in selected texts to “discover the themes that run through them, how those themes are connected to each other, and how they change over time” (Blei 2012). I have used the latent Dirichlet allocation (LDA) model, which is a probabilistic topic model that categorizes a set of documents into topics using probability distributions. As Blei, Ng, and Jordan (2003, p. 20) explain, an LDA model represents documents as “random mixtures over latent topics, where each topic is characterized by a distribution over words.” The topics are essentially themes of the document collection that summarize the collection (Chaney and Blei 2012). I have created topic models for each of the three cases. All coded data were analyzed using descriptive statistics. The results of the analysis are presented and discussed below.

89 The category of journalists has been included to account for any potential media slant. The category of experts has been added after the initial tests of the codebook to allow for a more fine-grained analysis of different framing actors.

3.3 Results and discussion

Tables 19 and 20 present the incidence of specific frame themes in the post-disaster news coverage of the disasters under study. Looking at the three cases together, the results are not surprising. Three types of frames tend to dominate the post-disaster discourse: blame assignment, economy, and environment. The topic models (provided in full in Appendix 4) support the selection of hand-coding categories as well as these initial results. Economy, environment, and blame assignment dominate the news after an environmental disaster. Other types of frames – morality and human interest especially – are much less prevalent in the post-disaster news coverage. Table 18 summarizes these three topics for each case, listing the most frequent and exclusive words.

Table 18. The prevailing topics in the news coverage after the Mount Polley, Deepwater Horizon, and Fukushima disasters.

Topic 1: Economy Mount Polley: industry, project, year, work, resource, development, jobs Deepwater Horizon: share, price, market, cost, billion, investor, dividend, shareholder, asset Fukushima: electricity, service, industry, company, manufacturing, work

Topic 2: Environment Mount Polley: water, lake, sediment, salmon, river, creek Deepwater Horizon: coast, water, surface, sea, area, fish, ocean, impact Fukushima: milk, vegetables, drinking water, foodstuffs

Topic 3: Blame assignment Mount Polley: panel, government, BC, design, report, investigation, review, province Deepwater Horizon: federal, agency, administration, investigation, government, Obama, report Fukushima: Japan, Germany, Merkel, politics, nuclear power station

As discussed earlier, the economy, environment, and blame assignment frames should, in theory, be linked to protest. Yet, since they are strongly present in all three cases, they do not easily explain the variation in post-disaster protests. For example, environmental frames have a much stronger presence in the Canadian case than in the other two. Yet, since the post-Mount Polley protest was minimal (as opposed to the other two cases), environmental frames may not be a significant motivating factor – at least not in the Canadian context. In Germany, however, environmental (as well as economy) frames were most prevalent after the Fukushima disaster. It is therefore unclear, at least at the general level, what role environmental frames (when present)

94 have in post-disaster protest mobilization. Similar reasoning applies to the incidence of economy and blame assignment frames.90 The industry frames were surprisingly absent in all three cases, taking up only a small percentage of the news coverage. This suggests that framing actors are, for some reason, not linking these environmental events to the bigger discourse of the environment and energy production.

Table 19. Incidence of frames in disaster news coverage (in percent, rounded).

Frame Mount Polley Deepwater Horizon Fukushima (N=183) (N=99) (N=206) Blame assignment 43 56 18 Blame denial 8 12 0.5 Human interest 14 2 12 Morality 4 3 13 Economy 26 35 30 Conflict 11 14 14 Environment 66 38 39 Pro-industry 4 1 5 Anti-industry 3 9 17 Neutral 7 5 4 Note: A newspaper article may contain any number of different frames. The above frame types are not mutually exclusive; although, most of the media coverage that I examined generally contained only one type of frames.

Given that general frame themes seem to shed little light on the post-disaster protest variation, it is useful to look at different aspects of these frames more closely. Environmental frames, for example, are likely to have mobilization potential, if they emphasize the scale of damage. However, sending competing messages (e.g., the damage is large vs. the damage is minimal) could have a dampening effect due to the resulting uncertainty.91 Similarly, economy frames are likely to have mobilizing potential, if they emphasize negative economic impacts, but may be less effective if they employ pro-industry narratives. Lastly, blame assignment frames are likely to be effective in mobilization, if there is only a very small number of factors to blame. Having too many scapegoats is likely to result in uncertainty, which may have a dampening

90 In Germany, the incidence of blame assignment after Fukushima was only 23%. Compared to the other two cases, this is rather low. It is, however, not surprising, since the German news media do not have an incentive to focus on blaming TEPCO for the disaster. Instead, they are more likely to pay attention to the implications of the disaster for German politics, economy, and society. 91 The potential effects of uncertainty on protest are discussed later in this chapter.

95 effect on mobilization. Table 20 presents a breakdown of the frame themes into a number of different framing variables and their respective incidence in the post-disaster news coverage. Below I focus on the three predominant themes: blame, economy, and environment. After the Mount Polley mine leak, the Canadian framing actors overwhelmingly assigned blame for the disaster to two actors – the BC Liberal Party in power at the time, and the Imperial Metals Corporation (the owner of Mount Polley) where the former was in a sharper focus. This dynamic was similar after the Deepwater Horizon oil spill. Although, in this case, multiple companies took part in blame assignment and deflection. In Germany, the blame assignment frames after Fukushima were somewhat different – mostly the focus was on the inadequacy of the existing practices when it comes to nuclear energy. Overall, there was little confusion about the direction of blame in all three cases. Therefore, blame frames alone do not explain the variation in post-disaster protest mobilization. The incidence of frames emphasizing the economic costs of the disaster was approximately the same for all cases. The German framing actors, however, focused on the economic consequences of environmental damage and on post-Fukushima policies about twice as much as the Canadian and American framing actors. Given that the post-Fukushima protests in Germany were the largest, economy frames may have a mobilization potential and should be investigated further. The environmental frames are not conflicting to a large extent in any of the cases. The predominant focus is on the scale of environmental damage, and the incidence of damage minimizing is very small. This suggests that, contrary to theoretical expectations, environmental frames are not a significant motivating factor in post-disaster protest mobilization. Although the incidence of the remaining framing variables is low, a closer look at them reveals several unexpected patterns. Much like the overall absence of industry frames, none of the disasters has been linked to broader environmental themes in order to, for example, call for climate action or the slowing of environmental degradation. Similarly, there has been no assignment of dramatic labels in any one of the cases, and the emphasis on dichotomies is very rare. This is in glaring contrast to public discourses surrounding climate change and environmental protection, and especially in the United States (for example, Dunlap, McCright, and Yarosh 2016; Bolsen and Shapiro 2018). Perhaps major environmental shocks like these disasters distract from the otherwise ongoing industry-environment discourses instead of serving as fuel or lightning rods. Lastly, the case comparison reveals that some framing variables that in theory should be significant in protest mobilization – human interest frames and narratives meant

96 to evoke emotions (specifically with respect to the environment and economy) – have the highest incidence in the case with the smallest size and number of post-disaster protests. This finding may have implications for established mobilization theories in general and environmental movements in particular.

Table 20. Percentage (rounded) of each framing variable.

Frame Percentage Mount Deepwater Fukushima Polley Horizon (N=183) (N=99) (N=206) Responsibility Some actor is responsible for the disaster. 25 37 2 Some other thing is responsible. 15 10 3 Government can alleviate the problem. 7 1 1 The existing practices are not working. 23 15 12 Accepting responsibility. 1 5 2 Employing excuses. 7 11 1 Employing justifications. 1 0 0 Disaster was ‘an act of God’, a natural 1 1 0 occurrence, or impossible to predict. Human Focus on affected individuals or groups. 24 6 11 interest Focus on personal or private lives of the 5 1 3 impacted individuals. Adjectives, metaphors or anecdotes that 2 1 8 generate feelings of fear, anger, empathy or sympathy. Focus on own personal experience or the 6 0 2 humankind. Morality Focus on unethical or immoral actions. 4 3 4 Social prescriptions. 2 2 0 References to morality or religious tenets. 0 0 1 Economy Economic costs of the disaster. 24 30 27 Economic consequences of the 5 5 12 environmental damage. Economic consequences of pursuing/not 2 4 8 pursuing a course of action. Adjectives, metaphors or anecdotes that 6 3 1 generate feelings of fear or anger linked to the economic consequences of disaster.

Frame Percentage Mount Deepwater Fukushima Polley Horizon (N=183) (N=99) (N=206) Conflict Disagreement among actors. 8 11 6 Actor criticizes or accuses another. 6 6 10 Assignment of dramatic labels. 0 0 0 Dichotomies. 3 2 1 Environment Environmental damage from disaster. 64 29 37 Disaster linked to broader themes. 0 3 1 Adjectives, metaphors or anecdotes that 26 17 8 generate feelings of fear or anger linked to the environment. Environmental damage is minimal. 14 4 2 Industry Benefits of the industry. 4 0 2 Stakeholder support for the industry. 0 1 2 Reliability or safety of the energy. 2 0 3 Risks of the industry independent of the 3 8 3 disaster. Need for alternative energy. 0 3 13 Other disasters caused by the industry. 2 1 3 Adjectives, metaphors or anecdotes that 2 3 4 generate feelings of fear or anger linked to the industry. Balanced view of the energy production. 0 0 2 Inevitability of the energy production. 1 0 2 Undecided or no position on the industry. 7 5 2 Trade-off between the energy and other 0 0 2 issues.

In addition to frame incidence, I have also analyzed the tone of post-disaster news coverage. Overall, the tone is mostly negative in all three cases. Again, tone in general terms therefore cannot explain the protest variation. A closer look at different types of negative tone is needed (see Table 21). In the Mount Polley and Deepwater Horizon news coverage, the predominant tone was linked to lack of preparedness and uncertainty surrounding the disaster. The highest incidence of any negative tone type was uncertainty – 27% after Fukushima. Here,

98 obscure tone had the highest incidence across the three cases, and so did the tone linked to relatability. The concept of relatability as a potentially mobilizing factor is linked to the idea of ‘communities of interest’, proposed by Birkland (1998, p. 54–55). Birkland has suggested that such communities may be thousands of kilometers away from the disaster, but their members identify with the harm that the disaster caused, believe that a similar disaster may occur in their own community, and fear such a possibility. Birkland has neither operationalized nor tested the idea, which in this study is captured through ‘relatability’– its potential effect will be discussed in a later chapter. The incidence of the uncertain tone as well as closely related tones (obscure and untrustworthy) is comparable between Mount Polley and Fukushima. This leads to two different conclusions: uncertainty is not the main protest mobilizing factor, or uncertainty has two opposing effects that likely manifest under different conditions. The latter alternative is theoretically more appealing. Psychological studies have found that individuals avoid taking action in response to uncertain information (Curley et al. 1986; Camerer and Weber 1992; Fox and Weber 2002). Therefore, larger uncertainty should be associated with a smaller pool of potential protesters. However, the Fukushima disaster was followed by a large-scale protest mobilization. Therefore, effects of uncertainty on protest (along with further theoretical propositions) will also be assessed through an experiment in Chapter 6.

Table 21. Tone of news coverage (incidence in percent, rounded).

Mount Polley Deepwater Horizon Fukushima Negative (N=183) (N=99) (N=206) Overall 53 48 61 Unsuccessful 9 12 18 Unprepared 25 19 3 Untrustworthy 18 6 18 Obscure 9 7 12 Uncertain 25 16 27 Relatable 5 10 18 Neutral Overall 32 42 31 Response 43 50 35 Management 7 9 1 Reliability 3 2 3 Information 2 0 3 Uncertainty 1 1 9 Relatability 0 0 2 Positive Overall 9 8 7 Successful 2 8 2 Prepared 2 1 1 Trustworthy 1 0 1 Informative 3 3 5 Certain 5 0 2 Unrelatable 3 0 2

Linking post-disaster frames to framing actors raises two questions. First, how well are specific actors covered in the news media? Second, of all statements that framing actors make in the news, which ones do they tend to use more often than others? Answering the first questions allows for a comparison of the prominence of actors in the news coverage. It, however, does not tell us what frames each actor is more or less likely to use. Answering the second question does just that. Table 22 below provides an overview of the news coverage of specific framing actors, while Table 23 shows the incidence of frames these actors used in the disaster aftermath.

100

Activists are featured very little in the post-disaster news coverage of Mount Polley, Deepwater Horizon, and Fukushima – 4%, 5%, and 3% of the time, respectively. This finding is in line with the literature on the protest paradigm, which is a pattern of media coverage characterized by the coverage in support of the status quo and the lack of coverage of “outsiders” (or those challenging the status quo) (Shoemaker 1984; Boyle, McLeod, and Armstrong 2012). In the aftermath of the Mount Polley disaster, of all statements that activists made, most contained blame assignment frames, followed by conflict (accusing) and environment frames. This is in line with theoretical expectations. After Deepwater Horizon, however, most activist frames were environmental. Here, the complete lack of blame assignment frames in activist statements is surprising. After Fukushima, 50% of all activist statements in the news contained conflict frames (accusing), followed by some blame assignment and morality frames (both 13% incidence) but no environment frames. Contrary to expectations, activists did not use human interest and economy frames at all after Deepwater Horizon and Fukushima, only after Mount Polley (7% incidence for both frame types). Governments’ use of frames in the post-disaster news coverage is more in line with theoretical expectations. In case of Mount Polley, of all their statements in the news media, local government assigned blame 22% of the time, and then focused on economy, environment, and human interest frames (due to aboriginal groups’ concerns about the impacts of the disaster on their communities). The provincial government, on the other hand, denied blame in 33% of their statements (the same percentage as their use of environment frames). Of all framing actors, the provincial government was the most likely target of accusations by other actors, which is not surprising, given the localized impacts of the disaster (i.e., on the province of British Columbia). The federal government, while featured very little in the news, focused predominantly on blame assignment. Similar dynamics between lower and higher levels of government is apparent in the Fukushima case. Most of the statements made by the local government focused on accusations, blame assignment, and anti-industry narratives. The federal government statements were more mixed, with most containing blame assignment, economy, and industry frames. The inter- governmental dynamics was not evident after the Deepwater Horizon spill – the federal government assigned blame 47% of the time, while other levels of government were not represented in the coverage at all. Statements from companies were also not well-covered in the post-disaster news. After Mount Polley and Fukushima, the companies were featured only 1% and 3% of the time,

101 respectively (compared to the 20% incidence after Deepwater Horizon). As expected, they focused mostly on blame assignment, economy, and pro-industry frames.92 Corporate statements were more critical of the industry after Fukushima than Mount Polley, which is reflective of the domestic energy politics (i.e., British Columbia’s reliance on the mining sector and Germany’s long-standing nuclear energy discourse). Overall, an examination of the links between framing actors and specific frame themes reveals some patterns that are both in line and contrary to theoretical expectations. Conflict and blame frames, for example, reflect the political dynamics between different strands of government, and, in case of the Deepwater Horizon, among multiple companies. Companies also tend to use pro-industry frames, while local governments and activists are linked to anti-industry narratives. Such findings are not surprising. The sparsity of activist framing in the post-disaster news is somewhat surprising. This finding raises questions about the role of mobilization elites in protest. Does it matter who mobilizes the public through framing? More specifically, if a mobilizing frame is present in the news coverage, does it matter at all who its author is? Some answers may be found through an experiment, which can also account for actors other than traditional mobilization elites (e.g., citizens or experts). The tone-actor analysis – the last component of this content analysis – largely met theoretical expectations (the full results are available in Appendix 5). Activist tone is negative in all three cases. The tone of company frames tends to be positive except for two variables – information provision and uncertainty (i.e., companies use obscure and uncertain tone). After Mount Polley, the local government’s tone was predominantly negative, which aligns with previous findings. The provincial government was mostly neutral or positive – exceptions are the use of negative tone when it comes to information provision and uncertainty. After both Deepwater Horizon and Fukushima, the federal government employed mostly negative tone. This points to the only significant difference in tone-actor dynamics across cases: the federal government’s use of negative tone is linked to cases of medium- and large-size protests. Again, the role of framing elites in protest mobilization comes to mind. Perhaps the alignment of government frames and protesters’ attitudes signals a possibility of success for protesters (with respect to the ability of protest to effect change). This, of course, is an established theoretical aspect of protest mobilization, but one that has not been closely examined in practice.

92 Interestingly, Imperial Metals also used a lot of environment frames, and attempted no blame denial at all.

102

Table 22. Percentage (rounded) of framing actors covered in the post-disaster news.

Framing Actor Mount Polley Deepwater Horizon Fukushima Journalist 37 42 62 Activist 4 5 3 Government (total) 20 12 14 Local 7 0 1 Provincial/state 11 0 0 Federal 2 12 13 Company 1 20 3 Expert 12 10 7 Other 13 7 5 Multiple 9 13 6

Table 23. Percentage (rounded) of frames out of all statements framing actors made in the news.

Journalist Activist Loc. Prov./ Fed. Company Expert Other Multiple gov state gov gov Mount (N=272) (N=30) (N=48) (N=96) (N=8) (N=7) (N=66) (N=79) (N=49) Polley Blame 11 36 22 5 50 25 67 26 13 assign. Blame 0 0 9 33 0 0 0 2 3 denial Human 13 7 13 0 0 0 2 7 0 interest Morality 1 7 9 3 0 0 0 7 0 Economy 20 7 17 10 17 25 0 15 10 Conflict 0 14 4 10 17 0 7 2 7 (accusing) Conflict 0 0 0 21 17 0 0 0 7 (accused) Environ. 46 14 17 33 17 25 21 35 57 Pro- 2 0 0 0 0 25 0 2 0 industry Anti- 1 7 9 0 0 0 0 0 0 industry Neutral 6 7 0 5 0 0 2 4 10

103

Deepwater (N=65) (N=7) (N=0) (N=0) (N=19) (N=31) (N=15) (N=10) (N=20) Horizon Blame 23 0 0 0 47 23 53 20 20 assign. Blame 0 0 0 0 0 59 0 0 0 denial Human 9 0 0 0 0 0 0 0 0 interest Morality 2 14 0 0 5 0 0 0 0 Economy 47 0 0 0 8 30 0 31 46 Conflict 2 0 0 0 5 10 13 0 5 (accusing) Conflict 0 0 0 0 5 10 0 0 10 (accused) Environ. 28 71 0 0 11 0 13 40 30 Pro- 0 0 0 0 0 0 0 10 0 industry Anti- 2 14 0 0 11 0 20 10 5 industry Neutral 5 0 0 0 11 0 0 0 0 Fukushima (N=188) (N=8) (N=4) (N=1) (N=39) (N=8) (N=20) (N=15) (N=19) Blame 10 13 25 0 13 0 20 27 11 assign. Blame 0 0 0 0 0 38 0 0 0 denial Human 13 0 0 0 0 0 5 0 0 interest Morality 3 13 0 0 0 0 15 0 0 Economy 26 0 0 0 13 0 15 7 16 Conflict 3 50 50 0 13 0 5 13 5 (accusing) Conflict 0 0 0 0 33 50 0 7 5 (accused) Environ. 37 0 0 0 3 13 5 7 32 Pro- 1 0 0 0 10 0 20 0 0 industry Anti- 6 25 25 100 13 13 0 40 26 industry Neutral 2 0 0 0 3 13 15 0 5

104

Conclusion

This chapter presented a descriptive study of the news coverage in the aftermath of three major environmental disasters: the Mount Polley mine leak, the Deepwater Horizon oil spill, and the Fukushima nuclear disaster. The prevailing studies of industrial environmental disasters tend to focus on only one aspect of post-disaster framing such as the blame dynamics or the role of media culture. In contrast, this chapter aimed to expand our understanding of how different framing actors talk about environmental disasters. To this end, the content analysis considered both the tone of coverage and different frame themes, including economic considerations, disaster impacts, moral and emotional appeals, and different industry positions. The results confirmed some theoretical expectations and defied others. The analysis reveals a surprising lack of frames commonly believed to play a significant role in protest mobilization – especially frames linked to emotions. Although emotional frames undoubtedly are important mobilizing factors in some contexts, they are neither emphasized nor seemingly necessary for protests after environmental disasters. Similarly, the role of environmental frames as mobilizing factors is debatable. All three disasters resulted in major environmental impacts, which were emphasized in the post-disaster news coverage regardless of the size of post-disaster protest. This suggests that environmental damage, although reported on, might be viewed as a loss of intrinsic rather than instrumental value (i.e., the damage is ‘out there’) – unless such damage is coupled with emphasis on relatability of the disasters (i.e., the damage affects us directly). The results from the tone-actor analysis lend some support to the mobilizing potential of relatability. Relatability is most prominent in the news coverage of Fukushima, followed by Deepwater Horizon, and Mount Polley. This roughly corresponds to the size of post-disaster protests in these cases. Lastly, this analysis also shows that contextual uncertainty is present after disasters, and especially in case of the largest protests, but a descriptive analysis cannot tell us what effect such uncertainty has on protest mobilization. This study therefore serves as a prelude to further research, and specifically an experiment described in the next chapter.

105

Chapter 6. Uncertainty Framing of Environmental Disasters and the Willingness to Protest

(co-authored with Eric Merkley)

Most framing literature explores framing as a strategic activity, focusing on the identification of frames that actors use to influence target audience. Similarly, the previous chapter explored different types of frames and tone that actors are more or less likely to use in the disaster aftermath. This chapter considers the receiving end: How do post-disaster frames influence target audience? Are they effective, and if so, why? The content analysis in Chapter 5 revealed that negative tone, and specifically one that emphasizes relatability and uncertainty of a disaster, is linked to larger post-disaster protest. The concept of relatability has not been explored in the literature, and there is no theoretical guidance in terms of expectations of its effects. For now, an in-depth exploration of relatability is therefore left to further research. When it comes to the effects of uncertainty, however, there seems to be a discrepancy between theoretical expectations and empirical findings. On one hand, psychological studies suggest that uncertainty has a dampening effect on individual willingness to take action (Curley et al. 1986; Camerer and Weber 1992; Tversky and Shafir 1992; Fox and Weber 2002). On the other hand, some highly uncertain disaster events have been followed by large-scale protest mobilization. The purpose of this chapter is therefore twofold: first, to explain the contradiction between theory and empirical reality, and second, to gain a better understanding of the effects of disaster communication on public willingness to participate in non-violent protest such as peaceful demonstrations, petitions or boycotts. We focus on post-disaster uncertainty as communicated to the public through news media. We rely on the findings of the text analysis of news media coverage of three major industrial environmental disasters linked to varying degrees of post-disaster protest: the Mount Polley mine leak, the Deepwater Horizon oil spill, and the Fukushima nuclear disaster (see Chapter 5). Drawing on the existing literature on framing and motivated reasoning, we propose four hypotheses related to the effects of uncertainty on people’s willingness to protest. We posit that individuals exposed to uncertainty framing are likely to become anxious and are less likely to participate in protest activities. This effect of uncertainty is likely to be dampened, if the communication of uncertainty suggests that the costs of the disaster may be higher than

106 expected. Lastly, depending on their ideological leaning, individuals might resist or give in to the demobilizing effect of uncertainty. To test our proposed theory, we conducted a survey experiment of 3,600 adults in the United States, representative of the broader US population. The participants were recruited through Amazon’s Mechanical Turk (MTurk), an online platform where workers find and complete tasks for modest payments. The results show that uncertainty framing had no effect on respondents’ emotions, including anxiety, anger or negative affect in general. This goes contrary to the psychological literature on the anxiety-inducing effect of uncertainty, as well as the work that argues for a distinction between different types of negative emotions and their effects (Johnson and Tversky 1983; Lerner and Keltner 2001; Lerner et al, 2003). Similarly, emphasis on possible high costs of the disaster had no effect in reducing the likelihood of protest. The last hypothesis, however, was strongly supported. The effectiveness of the uncertainty framing is conditional on political ideology, with more liberal respondents being more likely to resist such framing and therefore more likely to participate in a protest activity. This finding reveals a new interaction effect between uncertainty and political ideology, one that warrants further investigations. This chapter proceeds in four main parts. First, to ground the experiment in both the results of the content analysis and the prevailing literature, we provide a theoretical background on frame effectiveness and the role of negative frames (i.e., tone) in individual willingness to take political action. We then discuss our theoretical expectations, presenting four hypotheses on uncertainty, emotions, pre-existing beliefs, and the willingness to protest. The third part consists of the experimental design, data, analysis, and findings. Lastly, we end with a conclusion of findings and discussion of future research.

1. The Framing Theory Continued

While the previous chapter discussed the existing framing research in terms of what frames are, how they work, and how framing actors use them, this one focuses on the characteristics that make them work as intended. This section therefore explores the theoretical

107 underpinnings of two relevant issues: What makes frames effective?93 How do frames affect their audience, and particularly under the conditions of uncertainty that disasters generate?

1.1 Frame effectiveness and frame strength

A large body of literature has established that frames affect behavior in every aspect of life – from healthy living to formation of political attitudes to coalition bargaining to protest mobilization (for example, Quattrone and Tversky 1988; Druckman 2004). However, frames do not always lead to their intended outcomes. As discussed below, the prevailing literature on frame effectiveness has established some main ingredients of effective framing. However, from the content analysis in Chapter 5, it is apparent that these elements of frame effectiveness fail to fully explain different protest outcomes in the aftermath of environmental disasters. Many scholars believe that it is frame strength that ultimately makes frames effective. Based on an extensive literature review, Busby et al. note four main factors that determine frame strength: source (credible vs. non-credible), episodic focus (concrete experience vs. thematic focus), emotion, and identity threat (see also Haigh et al. 2006; Chong and Druckman 2007; Gross 2008; Aarøe 2011; Arceneaux 2012; Klar et al. 2012; Klar 2013). With respect to successful protest mobilization, scholars emphasize frame resonance, determined by credibility of frame94 and frame’s relative salience (Snow and Benford 1988 and 2000). Salient frames engage with values that the audience finds essential (Pastor et al. 2011; Chong 2000; Chong and Druckamn 2007); they are linked to receivers’ personal experiences and resonate with cultural narratives through the use of words or images that are “noticeable, understandable, memorable and emotionally charged” (Entman 2003). After a disaster, the public will likely receive multiple competing frames related to the event (Chong and Druckman 2007; Lim and Seo 2008; Klar et al. 2012). The strength and

93 Frame effectiveness can be understood in two ways. First, framing elites select those frames they believe will work – for example, because they have worked in the past (see, for example, Zald 1996, p. 261). Frames are then deemed effective even before they are (re)-used. Second, frame effectiveness is evaluated with respect to the occurrence of the intended outcome (i.e., if the intended outcome occurs, the frame is effective). This view of effectiveness accounts for unintended factors (e.g., some major other events occurs that captures people’s attention) that may shape the framing dynamics and even render frames that are generally considered effective unsuccessful. This research considers both elements of effectiveness. While the content analysis in Chapter 5 teased out frames effective in protest mobilization given the outcomes (i.e., protests) that already occurred, the experiment in this chapter assumes that those frames are likely effective even if re-used in the future (and under laboratory conditions). 94 Credibility involves the above-mentioned source credibility but also empirical credibility (i.e., the apparent fit between the frame and the real world – the frame needs to be empirically verifiable).

108 prevalence of a particular frame determines the overall effect of framing on the public (Lim and Seo 2008). Effective framing (by any political actor) will then involve frequent appearance of certain frames (either negative or positive) in form of culturally accepted symbols, metaphors or images (Cooper 2002; Entman 2003). The content analysis in the previous chapter identified three types of prevailing frames: blame assignment, economy, and environment. However, these as well as the less prevalent frames were mixed in terms of their strength. No one case was linked to significantly stronger or weaker frames than others. The sources were credible (strong), human interest and various emotional frames were relatively few (weak), and there were no identity threatening frames in any of the cases (weak). The one characteristic of post-disaster frames that stood out was their negative tone. The theoretical significance of such tone is discussed below.

1.2 Negative frames, Prospect Theory, and post-disaster protest

Negative frames (i.e., those that emphasize losses or disadvantages) have been shown to have more lasting effects than positive frames (i.e., those that emphasize successes or gains) (Ledgerwood and Boydstun 2014; Lecheler and de Vreese 2016). This finding comes from studies based on Tversky and Kahneman’s prospect theory. Prospect theory explains how (and why) individuals make choices under conditions of uncertainty – and how these choices are not always ‘rational’. The type of framing that prospect theory was applied to was ‘risky choice’ framing, because the descriptions of risk in each experimental treatment group varied.95 In their experiments, Tversky and Kahneman (1979, 1984) showed that people’s decisions depend on the ways in which an issue is described. The choices given to participants were logically equivalent in their expected value but different in the ways they were described. Perhaps the most famous is the so-called ‘Asian disease problem’, in which participants were given a choice between two strategies for addressing a disease: 1) the gain frame with two options: saving a fixed number of people (safe option) or a 1/3 chance that everyone is saved,

95 There are two more types of framing commonly compared to risky choice – attribute and goal framing (Levin, Schneider, and Gaeth 1998); however, it is debatable whether there are any significant differences between the two or whether they are simply different types of equivalence framing. Attribute framing (i.e., logically equivalent descriptions) is about “manipulating a single characteristic of a relevant event with a positive or negative frame” (e.g., quality of meat as 75% (positive) or 25% (negative) (Levin and Gaeth 1988). Goal framing changes the way in which the goal is described, but different messages recommend the same behavior to achieve a goal (i.e., positive = achieve desirable outcome; negative = avoid undesirable outcome through the same activity) (Krishnamurthy, Carter and Blair 2001; Maheswaran and Meyers-Levy 1990).

109 and a 2/3 chance that no one is saved (risky option), and 2) the loss frame with two options: a fixed number of people die (safe option) or 1/3 chance that no one dies and 2/3 chance that everyone dies (risky option) (Tversky and Kahneman 1981). When the scenario was presented in the gain frame, people, being more risk averse, preferred the fixed option and therefore the more certain payoff. When the scenario was framed in terms of losses, participants preferred the risky option. Prospect Theory therefore challenges the expected utility theory models, emphasizing, among other things, that expected utility theory does not explain how framing can change individual’s decision (Asgary and Levy 2009). Since people experience losses more intensely than gains, their motivation to minimize losses will be larger than their motivation to maximize gains of similar magnitude (Taylor 1991). Prospect theory is suitable in post-disaster protest mobilization for two reasons. First, disasters create conditions of uncertainty that shape the strategic environment in which framing actors operate and interact. Second, decisions related to protest participation represent risky choices – due to potential costs (and benefits) associated with protest action. Under such conditions, positive frames lead individuals to become risk averse, while negative frames prompt people to make riskier choices (Kahneman and Tversky 1979; Tversky and Kahneman 1981; 1986; 1991; 1992). In other words, people are more likely to act if subjected to negative rather than positive frames (McClure et al. 2009). Positive and negative frames bring to focus positive and negative consequences of one’s future decisions or behaviour. In disaster context, this has been applied in disaster planning, emergency preparedness and response (McLure 2009). The idea has not been applied widely in social movements literature and for protest mobilization in particular. Activists are expected to use frames that revolve around encouraging people to take action (e.g., to pressure government for stricter regulations) – this has partially been captured by the idea of diagnostic/prognostic frames, but, as noted earlier, the focus has been on the framing actors, not on frame receivers. Following the logic of the prospect theory, certain types of diagnostic/prognostic frames have a greater chance of success, because they make people less risk averse (and therefore willing to participate in protest). Applying this logic on disaster protest mobilization, if the prevailing tone of post-disaster coverage is positive, people should be less likely to participate in protests. In contrast, framing a disaster in negative ways should make individuals less risk averse and easier to be mobilized. The content analysis has shown that in all three cases the tone of

110 coverage was predominantly negative, yet, the post-disaster protests were different. This suggests it is the specific type of negative tone that may matter – and specifically uncertain and relatable. As noted earlier, this chapter only focuses on the former, leaving an examination of relatability to future research.

2. Towards a Better Understanding of Protest: Uncertainty and the Willingness to Protest

As discussed in Chapter 5, disasters generate uncertainty. Its effects on the framing dynamics and public willingness to protest are, however, unclear. Psychological studies have found that individuals avoid taking action in response to uncertain information (Curley et al. 1986; Camerer and Weber 1992; Tversky and Shafir 1992; Fox and Weber 2002). Therefore, in general, larger uncertainty should be associated with a smaller pool of potential protesters. However, as noted several times in the previous chapters, some highly uncertain events such as the Three Mile Island and Fukushima nuclear disasters were followed by a large-scale protest mobilization. Uncertainty therefore seems to have two opposing effects on protest in disaster aftermath – increase its likelihood and size or decrease its likelihood and size. Psychological literature suggests that high inherent uncertainty dampens protest mobilization. There are cognitive and emotional reasons. The first is largely due to cognitive effects of uncertainty on individual behaviour – the larger the uncertainty in communication the smaller the audience’s ability to understand and respond (Tversky and Shafir 1992). Second, uncertainty is a source of fear, which has been identified as a powerful anxiety-inducing factor (Greco and Roger 2003; Grupe and Nitschke 2013). Fear and anxiety produce pessimism about future outcomes and therefore risk aversion (Johnson and Tversky 1983; Lerner and Keltner 2000; Lerner and Keltner 2001; Lerner et al. 2004).96 This aligns with findings of social movements scholars who have identified the lack of possibility for success (i.e., absence of the feeling of ‘self-efficacy’) as a crucial factor that dampens people’s willingness to participate in political action (Kurzman 1996; Ruiter et al. 2001; Brown 2016). In the context of environmental disasters and frame effectiveness, this suggests the following two hypotheses:

96 Previous research has found that emotional states shape the impact of a given frame (Marcus et al. 2000; Brader 2006; Druckman and McDermott 2008; Arceneaux 2012). Anxiety and fear in particular make individuals more susceptible to frames, as people seek more information to cope with the stressful situation (Marcus et al. 2005; Weeks 2015).

111

H1: Individuals exposed to uncertainty framing are likely to feel more anxious than those not exposed to uncertainty framing.

H2: Individuals exposed to uncertainty framing are less willing to engage in post-disaster protest action.

The anxiety-inducing effect of uncertainty (and therefore risk aversion) may be dampened if the communication of uncertainty implies that the costs of the disaster may be higher than expected. Text analysis of news content (see Chapter 5) revealed that most stories of environmental disasters contain uncertainty frames about long-term environmental impacts (see Table 24). These frames tend to present the uncertainty as one in which problematic long-term effects may be worse than currently expected. Such communication is likely to invoke anger within both the affected communities and broader population. Experiencing anger has been found to trigger more optimistic beliefs about future outcomes, and encourage risk taking behavior (Lerner et al. 2003). In crisis communication, anger-inducing crisis news prompt individuals to form negative attitudes toward the actor at fault (e.g., the corporation responsible for a disaster) (Kim and Cameron 2011). Anger may therefore increase the willingness to protest after industrial disasters, counteracting the dampening effect of uncertainty. Therefore, we formulate the third hypothesis as follows:

H3: Negative effect of uncertainty on willingness to engage in protest action is lower when individuals are exposed to framing that implies possible high disaster costs.

Table 24. Uncertainty in the news media coverage of the Mount Polley, Deepwater Horizon, and Fukushima disasters (in percent).

Uncertainty type Mount Polley (N=47) Deepwater Horizon (N=17) Fukushima (N=57) Cause 30 24 2 Impact 77 77 97 Economic 4 18 2 Environmental 72 53 12 Upwards 75 77 78 Downwards 8 8 13 Pure/neutral 22 23 15

112

It is also possible that the effects of uncertainty are not contingent upon emotions but on individuals’ pre-existing beliefs, and particularly ideological leaning. We therefore add a hypothesis that stems from the psychological literature on motivated reasoning (see Kunda 1990; Ditto and Lopez 1992). To be effective, a frame needs a conducive audience. The ‘frame in communication’ must somewhat match the ‘frames in thought’ that people have about certain issues. Individuals’ ideological beliefs fundamentally shape the types of frames that could or could not be effective for them (see, for example, Slothuus and de Vreese 2010). For example, if people are ideologically motivated to resist or support an active government environmental policy, they will accept and reject frames accordingly (Kahan 2013). Under the conditions of uncertainty, people are more likely to interpret information in ways that align with prior beliefs and worldviews (Dieckmann et al. 2017; see also Kahan, Jenkins-Smith, and Braman 2011). Politically conservative individuals are resistant to change, prefer system stability, and are less tolerant of uncertainty and threat (Jost et al. 2003). Politically liberal individuals are more likely to protest than non-liberals (Schussman and Soule 2005). Therefore, we hypothesize that politically liberal individuals are motivated to resist the uncertainty frame, while politically conservative individuals are likely to accept it. The fourth hypothesis therefore follows:

H4: Effect of uncertainty frame is stronger with respondents with conservative political ideology.

A survey experiment, described in the following section, has been designed to test this and the other hypotheses.

3. Experiment

Experiments are one of the most persuasive ways of establishing causality. This section presents a randomized experiment to test the effects of specific post-disaster media frames on audience members. Randomized experimentation helps reduce the impact of unobserved variables that may affect the groups that receive a treatment versus those that do not. Such experiments are effective at isolating effects of the treatments to facilitate causal inference (Green and Gerber 2012). The administration of treatments through survey vignettes in particular

113 makes it possible to manipulate multiple independent experimental factors simultaneously (Mutz 2011). Such treatments can also mitigate some concerns of external validity, since they are sufficiently complex to better approximate the real-world conditions over more traditional laboratory settings (Mutz 2011). Below, we present the design, data, and findings from a survey experiment we have designed to test the effects of post-disaster uncertainty.

3.1 Design and methods

To test our hypotheses, we used a 3X2 experimental design. Our treatments were embedded in purposively designed mock news articles describing a fictional oil spill off the coast of California in May 2018. It was necessary to make the event novel to understand how public participation emerges from certain framing of disasters. If we had selected an event from the past, it was unlikely that any willingness to engage in protest action would have emerged, and, as a result, we would not have been able to test the proposed hypotheses. The article itself was designed to look like it was from the Reuters newswire service, but the bulk of the content came from a CNN article describing the Santa Barbara oil spill of 2015. This change was essential, so that participants would accept the information as credible. Citizens engage more with information when it is from sources they trust. Non-political newswire services are likely to be trusted by a wider range of Americans of different values and beliefs than a cable news network. The treatment articles were designed to vary in two ways: their use of uncertainty framing, and the presence or absence of a human safety element to the disaster. The primary dimension of the experimental manipulation was in the use of uncertainty frames. The first treatment group received a version of the article, in which the long-term environmental impacts of the oil spill were cast as fully uncertain, with a headline that read: “Long-term effects of oil spill on wildlife uncertain, officials say.” The framing included in the text of the article read: Both government officials and the researchers note that it is too soon to say whether or not local wildlife will be negatively affected by this spill in the long term. It is possible this variant of uncertainty framing does not have the same effect in suppressing protest mobilization as the framing with pure (or neutral) uncertainty, so the second treatment group received a slightly different variant of this article. The headline instead read: “Substantial long-term effects of oil spill on wildlife possible, officials say.” The text from above was slightly reworked to emphasize that the uncertainty is not about whether there will be a long-

114 term impact, but about how bad that impact will be: Both government officials and the researchers note that it is too soon to say how badly local wildlife will be affected by this spill in the long term. There is potential for the consequences to be much greater than anticipated. Subjects in the control group read a version of the article completely stripped of these statements of uncertainty. We also varied the existence of a human safety component to the disaster to ensure that the effect of uncertainty in depressing protest mobilization does not disappear when human health is jeopardized. In the control condition of this dimension, mention of human safety was, as above, omitted both from the headline and the in-text frame. The following statement was added to the article: There are not expected to be implications for human safety. Chemicals consumed by fish that are harvested and sold by local fisheries break down before being consumed by humans. In the treatment condition, the headlines and uncertainty statements reflected a possible effect on human health by the oil spill and included the following statement: There are also important implications for human safety. Fish consume these chemicals, which get passed on to humans who eat the seafood produced by local fisheries. These fisheries are an important staple of the community, as they sell their product to marketplaces throughout California. In all cases, the language added to the base article from CNN reflected actual language used in news coverage of environmental disasters. The treatment text can be found in Appendix 6, an example article in Figure 7, and our experimental conditions in Table 25 below.

Table 25. Treatment Conditions. Human Safety No Yes Uncertainty Framing Yes Yes/High Costs No

115

Figure 7. Example article – no uncertainty frames/no human safety.

116

Our general expectation is that uncertainty framing is associated with higher anxiety (H1) and reduced willingness to protest (H2), but that this will not apply when the frame is presented such that only the magnitude of the high long-term cost is in question (H3). We also expect the effect of uncertainty to be lower among the respondents on the political left (H4).

3.2 Data

Our experimental design has an advantage of being very realistic. The treatment frames are embedded in full-length newspaper articles, and reflect actual frames used in news content. This means, however, that we have a strong expectation that treatment effects will be small. Therefore, we conducted a power analysis to get a sense of what sample size we need to properly estimate any treatment effects that may emerge from our design. Our primary interest is in the comparison between the control condition with no human safety element and the first treatment condition with uncertainty framing but no human safety component. Our expectation is that our other manipulations may reduce the effectiveness of the frame. Therefore, we need a large enough sample to be able to precisely estimate this difference. We used the General Social Survey (GSS) to estimate the likely mean and standard deviation of an index of protest activities (re-scaled 0-1). In the GSS, the mean on this measure was 0.17 with a standard deviation of 0.21. We then used a power analysis to estimate our needed combined sample size across a range of possible mean values in the experimental condition at a conventional power of 0.8 and statistical significance at the 0.05 level. The results are shown in Figure 8. A combined sample size across the two conditions of 1,200 people would allow us to estimate an effect size of 0.16, which is reasonably small. The absence of a dampening effect by the human safety manipulation would also effectively double this sample size and allow us to estimate a treatment effect of 0.11 at a 0.05 significance level with 0.8 power. Thus, across all six conditions we decided to draw a combined sample of at least 3,600 respondents.

117

Figure 8. Power analysis results at a 0.05 significance level and 0.8 power.

We fielded the survey experiment in July 2018 to 3,628 adults in the United States who were recruited through Amazon’s Mechanical Turk (MTurk), an online platform for workers who consent to complete tasks (i.e., HITs) for modest payments. MTurk has been recognized and used by social scientists as a better means of acquiring samples for experiments than other convenience samples such as undergraduate students (Berinsky, Huber, and Lenz 2012; Mullinix et al., 2015; Hu and Tingley, 2015). Our sample cannot make claims to representativeness, but some of its broad characteristics are similar to the public as a whole. Table 26 provides a comparison of the 2016 GSS and the sample used in this research. The MTurk sample is reasonably representative of the American population in terms of gender, race, partisanship, and ideology, but it is substantially younger, more educated, less religious, and more affluent. This suggests our sample may tilt towards protest mobilization in response to an environmental disaster. The subjects may also be more inclined to resist our uncertainty framing than a representative sample of the general public. We have grounds to suspect our findings are conservative.

118

Table 26. Comparison of 2016 GSS survey and Amazon Mechanical Turk sample.

GSS (2016) MTurk Male 44% 42% White 73% 75% College Degree or Higher 30% 56% Conservative 34% 32% Republican (Lean Included) 35% 36% Monthly Church Attendance or Greater 44% 30% Employed Full-Time 46% 60% Under $20,000 Family Income 19% 12% Age (Mean) 49 39

We have two primary dependent variables of interest. First, we are interested in whether our uncertainty frames trigger an emotional response. Therefore, we have asked the following open-ended question after the treatment: “How did you feel after reading the newspaper article about the oil spill?” We then asked respondents to complete a battery of questions related to their emotional state, using an altered PANAS scale (see Watson et al. 1988; Kercher 1992). Our instructions were as follows: This scale consists of a number of words that describe different feelings and emotions. Read each item and then list the number from the scale below next to each word. Indicate to what extent you feel this way right now, that is, at the present moment. (Very slightly or not at all/A little/Moderately/Quite a bit/Extremely). The battery included the following emotional states: distressed, exited, upset, scared, enthusiastic, inspired, nervous, determined, alert, afraid, and frustrated. We had added ‘anxious’ and ‘angry’, since we believed these most closely reflect our motivating theory for H1. We also used a principle components analysis to determine whether there were any separable dimensions in these items. They appear to load on two distinct dimensions, which we call negative affect and positive affect. The factor loadings are shown in Appendix 6 (Table 40). Our second dependent variable is the willingness to protest. We included a battery of questions that asked respondents their likelihood of engaging in political activities in response to the oil spill. Our instructions were: Below are a number of political activities citizens engage in. In response to the oil spill you just read about, how likely are you to...” (Not at all likely/Somewhat likely/Likely/Very likely/Extremely likely). The activities they could engage in were: sign a petition; attend a lawful protest, rally or march; contact your local representative in Congress; donate to an environmental group; boycott companies with investments in the fossil fuel sector; look up more information on the oil spill or on related environmental issues; engage

119 in unlawful civil disobedience. We constructed a scale of protest activity from 0 to 1 that averaged these items. The experimental protocol is described below. After consenting to participate in the survey, respondents completed a background questionnaire on political values and basic socio-economic characteristics. They were then randomly assigned to one of six experimental conditions (approximately 600 respondents per condition) and asked to read the corresponding news article. After, they were asked to give their thoughts on what they read and filled out the batteries relating to their emotions and their willingness to engage in political activities in response to the oil spill. Since this experiment involved deception, a debriefing form was included at the end of the survey for participants to re- new their consent for participation and use of their responses in this research (see Appendix 6).97

3.3. Results

The results of difference of means tests for H1 through to H3 are shown below in Table 27. Our uncertainty framing did not have a meaningful or even significant effect on respondent emotions. It is non-significant for anxiety (p~0.43), anger (p~0.46), or negative affect more generally (p<0.22). There is little support for H1. Our uncertainty framing had no unconditional effect on the willingness to protest (p<0.34), so H2 is not supported. The type of uncertainty frame also appeared to not matter. Uncertainty frames that leave long-term costs unknown are no more effective in reducing the willingness to protest (p~0.37) than those that emphasize higher possible costs (p~0.44). There is effectively no difference between these conditions (p~0.90), which means no support for H3.

97 We also asked a pair of attention check questions throughout the survey, and 83% of respondents passed both. Results were not conditional on screener passage. Diagnostic tests are available upon request.

120

Table 27. Difference of Means Tests, H1-H3.

H1 H2 H3 Mean Difference Anxiety Anger Negative Protest Frame Type Uncertainty – Control 0.032 0.029 0.043 -0.008 (0.040) (0.039) (0.035) (0.008) Unknown Cost – Control -0.008 (0.009) High Cost – Control -0.007 (0.010) Unknown Cost – High Cost -0.001 (0.009) Standard errors in parentheses, *p<0.1, **p<0.05, ***p<0.01

There is, however, strong support for H4. It appears that the effectiveness of the uncertainty framing is conditional on political ideology. More liberal respondents are more likely to resist the framing of the treatment article. Table 28 presents the results of an Ordinary Least Squares estimation where our treatment is interacted with political ideology to predict one’s willingness to protest. Model 1 provides the results without controls. Every point on the 7-point ideology scale increases the treatment effect by 0.01 points such that movement across the entire range increases the dampening power of the uncertainty frame. We expect a reduction in the willingness to protest of 0.07 points or 0.30 standard deviations compared to someone who rates themselves as ‘extremely liberal’.98 Moderating variables are observational, so we need to control for other factors that may be associated with the willingness to protest and ideology. Model 2 includes interactions of the treatment with partisanship, political interest, and church attendance. The interaction effect holds up. The marginal effects from this model are plotted in Figure 9.

98 Since most people are clustered by the zero, 0.07 points is a considerable effect size.

121

Table 28. OLS Regression Estimates, H4.

H4 1 2 Uncertainty Frame 0.027 0.025 (0.017) -0.039 Ideology -0.038*** -0.039*** (0.003) (0.005) Uncertainty * Ideology -0.010** -0.016** (0.004) (0.007) PID -0.004 (0.004) Uncertainty * PID 0.007 (0.006) Political Interest 0.047*** (0.009) Uncertainty * Political Interest 0.004 (0.011) Church Attendance 0.017*** (0.002) Uncertainty * Church -0.003 (0.003) Constant 0.463*** 0.336*** R 0.12 0.17 N 3594 3387 Standard errors in parentheses, *p<0.1, **p<0.05, ***p<0.01

122

Figure 9. Effect of uncertainty frame on a willingness to protest conditional on political ideology.

These results are robust in other ways as well. There are significant interaction effects when replacing our willingness to protest index with its individual components on five of seven items. The type of uncertainty is also not a determining factor. These robustness tests are shown below in Table 29. The presence of uncertainty frames in environmental disasters appears to push some respondents back to the status quo and dampen their willingness to protest, but they must be ideologically predisposed to not resist such framing. In short, the results provide compelling evidence for H4.

123

Table 29. Robustness tests, OLS estimates.

Contact Look up Unlawful By Petition Rally Donate Boycott MC Info Act Type 1 2 3 4 5 6 7 8 Uncertainty 0.10 0.14 0.08 0.13 0.06 0.15 0.09 Frame (0.10) (0.09) (0.10) (0.09) (0.10) (0.10) (0.08) Ideology -0.23*** -0.18*** -0.13*** -0.14*** -0.23*** -0.15*** -0.01 (0.02) (0.02) (0.02) (0.02) (0.02) (0.02) (0.02) Uncertainty -0.04* -0.04* -0.04 -0.04* -0.03 -0.05** -0.04** * Ideology (0.02) (0.02) (0.02) (0.02) (0.02) (0.02) (0.02) Unknown 0.02 Cost (0.02) Unknown Cost * -0.01* Ideology (0.00) High Cost 0.03 (0.02) High Cost * 0.01** Ideology (0.00) 0.46** 2.70*** 1.55*** 1.63*** 1.90*** 2.09*** 2.56*** 0.55*** Constant * R 0.11 0.10 0.05 0.06 0.12 0.07 0.01 0.12 N 3596 3596 3596 3596 3596 3596 3595 3594 Standard errors in parentheses, *p<0.1, **p<0.05, ***p<0.01

Conclusion

Under what conditions do frames work (i.e., achieve the framing actors’ goal)? Or, in other words, when are they effective? The purpose of the experiment presented in this chapter was to gain an understanding of the effects of disaster communication on public willingness to participate in non-violent protest. We specifically explored the effects of uncertainty, since it is one of the defining characteristics of any environmental disaster, and one that poses an empirical puzzle. In theory, uncertainty makes people more risk averse and therefore less willing to participate in protest action, but large-scale protests have occurred even after highly uncertain disaster events.

While the psychological literature maintains that uncertainty has a dampening effect on individual willingness to partake in risky actions, such literature (and the experiments it is based on) is typically developed outside of political context. We therefore used the findings from the text analysis of actual newspaper content to specify different types of uncertainty that are likely to be used in disaster communication. We then developed and tested four hypotheses on uncertainty and public willingness to protest. These hypotheses were developed based on our theoretical understanding of the links between uncertainty, negative emotions (i.e., fear and anger) and the role of pre-existing beliefs in individual preference formation and resistance to frames. We fielded our survey to 3,600 adults across the United States, using Amazon’s MTurk. The findings reveal that uncertainty framing in the context of an environmental disaster has no effect on individuals’ emotions such as fear and anger. This suggests that emotions may not motivate people or lower their willingness to participate in protest activities. While the occurrence of a disaster generates uncertainty and anxiety, rather than these states actively influencing people’s willingness to protest, they serve as an ‘excuse’ for individuals to keep to their pre-existing beliefs. This challenges the psychological literature on the anxiety-inducing effect of uncertainty, as well as the work that argues for a distinction between different types of negative emotions and their effects (Lerner and Keltner 2001; Lerner et al, 2003). The effectiveness of the uncertainty framing is conditional on political ideology, with more liberal respondents being more likely to resist such framing and therefore more likely to participate in a protest activity. Our findings reveal that uncertainty of disaster impacts may be a critical element in people’s willingness to protest after environmental disasters. An examination of its interactions with other variables may offer further insights into the protest mobilization process. We evaluated how frames influence self-reported (i.e.. potential) behaviour. Without other information, these findings cannot be used to predict what will happen after a particular disaster – that will in part depend on the types of frames that dominate in the disaster aftermath. A disaster shrouded in uncertainty (and reflected as such in the media coverage) is likely to give rise to a narrower, more ideologically homogeneous protest coalition than if there is little to no uncertainty. These results could help us explain why some protest coalitions may have more breadth than others after disasters.

125

Chapter 7. Conclusion

Frequently, environmental industrial disasters attract little public attention, and motivate few individuals to participate in a political action such as protest. When post-disaster protests do occur, some spiral into large movements, but many merely flicker and subside. This dissertation explores this empirical variation in post-disaster protest movements in advanced industrial democracies. Through an examination of 38 cases of environmental industrial disasters that occurred over the past six decades, I expand the current understanding of protest emergence and growth as well as the social effects of environmental disasters in general. Due to the number and diversity of my research questions, I adopt a mixed-methods approach, relying on four different methods and tools: a geographic information systems analysis, qualitative comparative analysis, content analysis, and survey experiment. Overall, my research adds several new layers to the study of social movements and disasters. First, environmental disasters are geospatial phenomena, meaning that various types of data associated with these events are tied to locational information. Disasters represent specific types of environmental-societal dynamics, and often result from interactions among different environmental, social, political, economic, and cultural processes. Such processes cannot be studied in isolation from the natural environment in which they are embedded. Therefore, an understanding of disaster effects on human societies requires a consideration of both socio- political and geographical perspectives. It is for this reason that I have decided to conduct a GIS analysis to answer my questions about geospatial characteristics of disasters (and specifically the damage they cause) and their political impacts. To my best knowledge, my GIS analysis is the first comprehensive geospatial analysis of large industrial disasters and their social effects to date. While obtaining accurate and up-to-date data is time-consuming and often expensive,99 conducting a GIS analysis is an excellent way to analyze and visualize disaster impact data that would otherwise be difficult to process. My efforts allowed me to formulate new theoretical propositions about disaster damage and associated public willingness to protest, as well as suggest several areas for future research. For example, I argue that the concept of grievances – and especially as they relate to environmental concerns – needs further theoretical refining.

99 For example, finding accurate locations of nuclear power plants and detailed documentation of damage from nuclear disaster events has been especially challenging.

126

Second, I conducted a qualitative comparative analysis to test some empirical assumptions about post-disaster protest. Given my mid-sized dataset, QCA was the most suitable method of analysis, as it has been designed (but not restricted) to medium-N research (see Ragin 2008). The most prominent challenge of using QCA in my work was data collection. My original dataset, as presented in Chapter 3, contains cases of disasters from the mid-20th century and onwards. Some types of data, such as data on income inequality and industrial production, were not available before certain time periods. Other types of data, such as information on ongoing activist campaigns or human rights abuses, were incomplete or inaccessible due to paywalls. This unavailability and inaccessibility of data forced me to narrow my original dataset from 38 to 16 cases of disaster events, which limits the generalizability of my findings to the post-1992 era. While I confirmed many of the expected structural conditions theorized to be essential to enable or prevent protest mobilization, I found only one necessary condition for protest emergence, and discovered some new interactions of factors worth investigating further. Third, I added uncertainty as a crucial element in the post-disaster protest mobilization process, but one that can and should be included in analyses of movements after critical events in general. Such events, including major environmental disasters, generate uncertainty, which is then communicated and shaped by different political actors. In my research, I studied post- disaster communication and specific types of media frames from several thousand newspaper articles across three countries and two languages. My goal was to better understand how governments, corporations, and activists talk about disasters and what effects such language has on public willingness to protest. A combination of a newspaper text analysis and an experiment was the most fitting methodological approach. While the text analysis revealed the types of frames that framing actors actually use, the experiment helped establish whether these frames have any effect. Much of the social movements research focuses on the use of frames for specific purposes, leaving frame effectiveness aside. However, understanding whether, how and why frames work is critical to our understanding of their creation and use. In my research, I propose a new mechanism through which the uncertainty frames may affect public willingness to protest – the pre-existing political ideology as opposed to emotions (which have been the focus of much of the recent literature). Due to practical constraints, the survey sample was drawn from the US population only. This, again, may limit the generalizability of my findings. Furthermore, more

127 research is needed to explore my proposed mechanism in depth. For example, the findings from the experiment can be applied in a process-tracing study where the causal mechanisms can be further refined and tested. The generalizability of my overall findings can be evaluated in at least two ways. First, the role of sudden grievances, inequality and uncertainty may be further examined in the context of natural disasters and other contingencies. Second, the study’s scope conditions may be altered or expanded – for example, by focusing on non-democratic states or non-Western democracies. This would allow for an examination of conditions that are characteristic of rich democracies to see whether the direction of their effect holds. What specific effects these factors may have in different contexts is left for further research. Several of my findings suggest more specific areas for new research. For example, further application of GIS tools and survey experiments would help answer some of the questions that arise from this research. Due to difficulties in collecting geospatial data and financial constraints of conducting survey experiments, I was not able to answer these questions here. Below, however, I discuss some theoretical propositions that can be taken up by others. The GIS analysis in Chapter 3, for example, poses several new questions. In order to compel us to action, disasters have to impact what we care about but from instrumental rather than moral perspective. Why does an intrinsic value of nature not serve as an effective motivation for political action? Perhaps this is as much an ethical or philosophical query as one that should be posed to scholars of social movements. My mapping of disasters and protests led to further questions: What is the relationship between the proximity of populated areas to nuclear power plant facilities (or other potentially dangerous or polluting sites) and public willingness to protest after environmental disasters? Do cumulative effects of repeated disasters in the same area increase or decrease public willingness to protest? What is the role of local dependence on the responsible industry in post-disaster protest mobilization? The last two questions in particular give rise to two sets of competing hypotheses. First, while cumulative effects are an important aspect of corporate and government environmental assessments, their social impacts are not well understood. Theoretically, repeat disasters could have two opposing social effects: desensitization and therefore decreased willingness to protest, or public outrage over environmental destruction and therefore increased willingness to protest. A combination of a GIS analysis and a case study work such as process-

128 tracing could provide evidence in favour of one or the other. Such methodological approach might also be useful in shedding light on the second set of competing hypotheses, related to the role of local population in post-disaster protest emergence. After a disaster, the population of the impacted area may grow angry (and therefore more willing to protest), especially if their lives or livelihoods were significantly affected. However, if a large portion of this population is dependent on the responsible industry (e.g., through employment), they may be less willing to participate in a protest action against that industry. Testing these competing hypotheses would improve our understanding of the interactions between local economies and environmental public goods. My research also explores the use and impact of media frames. I questioned the significance of emotional and dramatic frames in protest mobilization, suggesting that emotional frames may not be as effective as previously thought – or at least not equally effective across contexts. Furthermore, the sparse presence of activist framing regardless of protest size raises questions about the role of elites in framing and protest mobilization. What are the links between frame effectiveness and the actors who use those frames? Is frame effectiveness fully or partially dependent upon the framing actor? If so, why? One may begin answering such questions through experimental methods. Another topic for future experimentation arises from my content analysis, which suggested that two types of negative frames – uncertainty and relatability – may play some role in protest mobilization. However, as a descriptive exercise, the analysis was unable to explain what this role is. Uncertainty was explored in Chapter 6, but relatability was left for further research. Here, I want to lay some theoretical groundwork for this future work. The anxiety-inducing effect of uncertainty (and therefore risk aversion) may be dampened in presence of relatability (or a belief that a similar disaster may happen nearby, sometimes in the future). Relatability is about making an event personal, and a low degree of relatability has been linked to lesser public concern about impacts of phenomena such as climate change (Leiserowitz 2005). From an individual perspective, the degree of relatability likely depends on a combination of prior experiences and beliefs, and the ‘relatable’ frames an individual is exposed to. With respect to the latter, relatability has at least two dimensions – an event can be made relatable through human interest frames (such as emotional stories about affected individuals) or through comparisons to the similar past or potential future events in one’s country, region or community.

129

Relatability is therefore conveyed through frames that evoke emotional response. Human interest frames may elicit feelings of sympathy, pity or anger, depending on the content of a given frame (Cho and Gower 2006; Gross 2008). If such frames emphasize injustices – for example, the disproportional impact of disaster on marginalized communities – they are likely to provoke anger, and, as discussed in Chapter 6, encourage risk taking behavior and the willingness to protest. The second way of establishing relatability is through comparisons of the disaster to other similar events – whether they occurred in the past or may occur in the future. Such comparisons could evoke either anger or fear. If the relatability frames emphasize cumulative impacts (i.e., the impacted area has either been damaged in the past or is likely to be damaged in the future), they are likely to elicit feelings of anger. Communities located near major oil tanker routes or pipelines, for example, are more prone to repeat oil spills, which may generate (and increase) grievances among the affected population. If the relatability frames focus on some major past disasters (such as Chernobyl in the case of the Fukushima disaster), the resulting emotion is likely to be fear. This fear of future impacts, however, is likely distinct from the fear of immediate disaster impacts, linked to individual concerns for safety and property. With more distant time horizons, individual concerns are likely to shift from health or property damage to worries about the future. Research in social psychology, and specifically on future discounting, suggests that when the future is unpredictable or survival uncertain, people tend to devalue the future and engage in more risk- taking in the present. Such findings, however, are linked to biological and developmental factors, and focus on socially disapproved risky behaviour with unwanted outcomes such as alcoholism and adolescent delinquency (Hill, Ross, and Low 1997; Hill, Jenkins, and Farmer 2006). The literature on the logic and mechanism of positive risk-taking behavior (with wanted outcomes such as beneficial policies or environmental conservation) has not yet been developed. In such cases, risk-taking behavior may be tied to perceived future losses and follow the logic of prospect theory. The fear of future impact may therefore dampen the anxiety-inducing effect of post-disaster uncertainty and increase the willingness to protest. This research focuses on protest rather than the ability of protest to affect change. Nonviolent protest embodies potential for political and social changes such as enactment of new laws and regulations, institutional transformations, and even shifts in culture and societal

130 perceptions. To examine such changes that may result from post-disaster protest activities, one would need to consider additional conditions such as pre-existing cultures. The concept of environmental culture in particular needs more attention. Regulatory environments, for example, are likely linked to environmental grievances. The number and type of relevant regulations as well as regulatory performance not only affects resilience building in societies but may also encourage or discourage people from protesting after environmental disasters. Well-performing regulations may increase resilience to future disasters, reduce the intensity of cumulative disaster impacts, and therefore lessen environmental grievances through signaling some level of control over future disaster impacts. In part due to lack of data availability, it was not in the scope of this research to evaluate regulatory environment or resilience and vulnerability in all the cases. More in-depth qualitative case research is necessary to tease out case-specific conditions for post- disaster protest, and especially those related to culture and risk perceptions. For example, established risk perception literature shows that people tend to be more concerned with topics that are ‘affectively loaded’ (Slovic 2000). There may be a link between such topics and pre- existing movements and relevant prior contentions. Given the increasingly polarized and disillusioned public (at least in Western societies), the pervasive presence of uncertainty in political life, and the growing urgency of environmental problems, studying the mechanisms and outlets for public discontent is vital for well-functioning democracies. Understanding how and under what conditions major environmental disasters and other contingencies encourage or dampen protest mobilization opens opportunities for peaceful resolution of social conflict and easing of social discontent. In my future research, I plan to further investigate the effects of disaster uncertainty on the public. I will focus on the strategic use of uncertainty by different political elites, examining the role of misinformation in disaster risk communication, and specifically the use of misinformation by political actors to frame environmental disasters for some political purpose. What is the role of misinformation in disaster planning? What is the extent of disaster denial and various types of cover-up in democratic states? What are the reasons for such practice, and what effects do fabrication or cover-ups of disasters have on disaster preparedness at individual, community, and institutional levels? The literature on the topic of disaster misinformation is sparse, and largely limited to information sharing during crisis (Faulkner 2001; Sutton 2010; Huang et al. 2015) and risk communication linked to climate change (Bailey, Giangola, and

131

Boykoff 2014; Bertolotti and Cattelani 2014). Furthermore, many of the existing case studies are either dated or not easily applicable to modern democracies. My goal is to better understand how political elites shape or create post-disaster uncertainty (e.g., through disinformation), and what effects these practices may have on social discontent, political behavior, and disaster risk reduction policies. The potential negative impacts of disaster disinformation are profound. Such practices may lead to lack of disaster preparedness and reduced resilience. In addition, presenting people with false information, and sowing doubt and uncertainty may deepen political polarization, with impacts on well-functioning democracy. With increasing risks of future disasters (IPCC 2014), better preparedness is crucial, and it becomes critical to understand how elite interests and fragmented media landscape shape the public understanding of disaster risks, how vulnerable the public is to disaster misinformation, and what potential consequences such vulnerability has on individuals, communities, and states.

132

Bibliography

Acemoglu, Daron and James Robinson. 2006. Economic Origins of Dictatorship and Democracy. Cambridge, UK: Cambridge University Press.

Ackerman, Peter. 2007. “Skills or Conditions: Which Key Factors Shape the Success or Failure of Civil Resistance?” Paper presented at conference on Civil Resistance and Power Politics, St. Antony’s College, University of Oxford, Oxford, UK, March 15-18.

Adamowicz, Wiktor, Peter Boxall, M. Williliams, and J. Louviere. 1998. “Stated Preference Approaches for Measuring Passive Use Values: Choice Experiments and Contingent Valuation.” American Journal of Agricultural Economics 80: 64–75.

Adeola, Francis O. 2011. Hazardous Wastes, Industrial Disasters, and Environmental Health Risks. New York: Palgrave Macmillan.

Ad Hoc Commission of Inquiry. “The Buffalo Creek Flood and Disaster: Official Report from the Governor's Ad Hoc Commission of Inquiry 1973.” West Virginia Archives and History. http://www.wvculture.org/history/disasters/buffcreekgovreport.html.

Aigner, Erin, Joe Burgess, Shan Carter, Joanne Nurse, Haeyoun Park, Amy Schoenfeld, and Archie Tse. 2010. “Tracking the Oil Spill in the Gulf.” The New York Times, 2 August. http://www.nytimes.com/interactive/2010/05/01/us/20100501-oil-spill- tracker.html?_r=1&.

Albala-Bertrand, J.M. 1993. Political Economy of Large Natural Disasters: With Special Reference to Developing Countries. Oxford: Clarendon Press.

Aldhous, Peter and Zena Iovino. 2011. “Japan’s Record of Nuclear Cover-ups and Accidents.” New Scientist, 18 March. https://www.newscientist.com/article/dn20263-japans-record- of-nuclear-cover-ups-and-accidents/.

Aldrich, Daniel P. 2008. Site Fights: Divisive Facilities and Civil Society in Japan and the West. New York: Cornell University Press.

———. 2014. “The Emergence of Civil Society: Networks in Disasters, Mitigation, and Recovery.” In Risk Governance, edited by Urbano Fra.Paleo, 135–148. Dordrecht: Springer.

Alexander, D.A. 2005. “An Interpretation of Disaster in Terms of Changes in Culture, Society and International Relations.” In What Is a Disaster: New Answers to Old Questions, edited by R.W. Perry and E.L. Quarantelli, 25–38. Philadelphia: Xlibris.

133

Allen, Henry. 1979. “May Days for the No-Nukers; A New March, Another Cause - And Echoes From the Past,” The Washington Post, 3 May.

Alvarez, R. Michael and Charles H. Franklin. 1994. “Uncertainty and Political Perceptions.” Journal of Politics 56: 671–688.

Alvarez, Sergio, Sherry L. Larkin, John C. Whitehead, and Tim Haab. 2014. “A Revealed Preference Approach to Valuing Non-Market Recreational Fishing Losses from the Deepwater Horizon Oil Spill.” Journal of Environmental Management 145: 199–209.

Amenta, Edwin, Neal Caren, Elizabeth Chiarello, and Yang Su. 2010. “The Political Consequences of Social Movements.” Annual Review of Sociology 36: 287–307.

Anderson, Alison and Agnes Marhadour. 2007. “Slick PR? The Media Politics of the Prestige Oil Spill.” Science Communication 29 (1): 96–115.

Arceneaux, Kevin. 2012. “Cognitive Biases and the Strength of Political Arguments.” American Journal of Political Science 56 (2): 271–285.

Armour, Audrey, ed. 1984. The Not-In-My-Backyard Syndrome. Downsview, ON: York University Press.

Arnold, Lorna. 1995. Windscale 1957: Anatomy of a Nuclear Accident. London: Macmillan.

Arthur, B. A. 1989. “Increasing Returns, Competing Technologies and Lock-in by Historical Small Events: The Dynamics of Allocation under Increasing Returns to Scale.” Economic Journal 99: 116–131.

Asgary, Ali and Jack Levy. 2009. “A Review of the Implications of Prospect Theory for Natural Hazards and Disaster Planning.” International Journal of Environmental Research 3 (3): 379–394.

Badruddin. 2003. Global Peace and Anti-Nuclear Movements. New Delhi: Mittal Publications.

Bailey, Adriana, Lorine Giangola, and Maxwell T. Boykoff. 2014. “How Grammatical Choice Shapes Media Representations of Climate (un) Certainty.” Environmental Communication 8 (2): 197–215.

Balseiro, C.F., P. Carracedo, B. Gomez, P.C. Leitao, P. Montero, L. Naranjo, E. Penabad, and V. Perez-Munuzuri. 2003. “Tracking the Prestige Oil Spill: An Operational Experience in Simulation at MeteoGalicia.” Weather 58: 452–458.

134

Barton, Allen H. 1963. Social Organization under Stress. Washington, DC: National Research Council, National Academy of Sciences.

———. 1969. Communities in Disasters: A Sociological Analysis of Collective Stress Situations. Garden City, NY: Doubleday.

Barrington, Lowell. 2012. Comparative Politics: Structures and Choices. Boston, MA: Wadsworth.

Battist, Lewis and Harold T. Peterson, Jr. Radiological Consequences of the Three Mile Island Accident. Washington, D.C.: Office of Standards Development, U.S. Nuclear Regulatory Commission.

Batty, Michael. 2003. “Using Geographical Information Systems.” In Key Methods in Geography, edited by Nicholas J. Clifford and Gill Valentine, 409–423. London: SAGE Publications.

BBC. 2000. “Death of a River.” BBC News, 15 February. http://news.bbc.co.uk/2/hi/europe/ 642880.stm.

———. 2008. “Chinese Landslide Toll Passes 250.” BBC News, 13 September 2008. http://news.bbc.co.uk/2/hi/asia-pacific/7614653.stm.

———. 2011. “Japan PM Says Fukushima Nuclear Site Finally Stabilized.” BBC News, 16 December. http://www.bbc.co.uk/news/world-asia-16212057.

Beardsley, Kyle and Brian McQuinn. 2009. “Rebel Groups as Predatory Organizations: The Political Effects of the 2004 Tsunami in Indonesia and Sri Lanka.” The Journal of Conflict Resolution 53 (4): 624–645.

Beech, Hannah. 1999. “How Japan's Accident Stacks Up Next to Chernobyl.” Time, 11 October. http://content.time.com/time/world/article/0,8599,2053740,00.html.

Benford, Robert D. and David A. Snow. 2000. “Framing Processes and Social Movements: An Overview and Assessment.” Annual Review of Sociology 26: 611–639.

Bennett, Nelson. 2016. “Mount Polley Disaster Revisited: It Could Have Been Worse.” Business Vancouver, 16 August. https://www.biv.com/article/2016/8/mount-polley-disaster- revisited-it-could-have-been/.

Bennett, W. L. 1980. Public Opinion in American Politics. New York, NY: Harcourt.

Benthall, Jonathan. 1995. Disasters, Relief, and the Media. 2nd ed. New York: I.B. Tauri.

135

Berkowitz, L. and E. Harmon-Jones. 2004. “Toward an Understanding of the Determinants of Anger.” Emotion 4: 107–130.

Berinsky, Adam, Gregory Huber, and Gabriel Lenz. 2012. “Evaluating Online Labor Markets for Experimental Research: Amazon.com’s Mechanical Turk.” Political Analysis 20 (3): 351–368.

Berkowitz, Leonard and E. Harmon-Jones. 2004. “More Thoughts about Anger Determinants.” Emotion 4: 151–155.

Bertolotti, Mauro and Patrizia Catellani. 2014. “Effects of Message Framing in Policy Communication on Climate Change.” European Journal of Social Psychology 44 (5): 474–486.

Birchfield, Vicki. 2008. Income Inequality in Capitalist Democracies: The Interplay of Values and Institutions. University Park, PA: The Pennsylvania State University Press.

Birchwater, Sage. 2014. “Likely Residents Fear the Worst after Mount Polley Mine Disaster.” The Williams Lake Tribune, 5 August. http://www.wltribune.com/news/likely-residents- fear-the-worst-after-mount-polley-mine-disaster/.

Birkland, Thomas. 1997. After Disaster: Agenda Setting, Public Policy, and Focusing Events. Washington, DC: Georgetown University Press.

______. 1998. “Focusing Events, Mobilization, and Agenda Setting.” Journal of Public Policy 18 (1): 53–74.

Birkmann, J., P. Buckle, J. Jaeger, M. Pelling, N. Setiadi, M. Garschagen, N. Fernando, and J. Kropp. 2010. “Extreme Events and Disasters: A Window of Opportunity for Change? Analysis of Organizational, Institutional and Political Changes, Formal and Informal Responses after Mega-disasters.” Natural Hazards 55: 637–655.

Bishop, Bradford H. 2014. “Focusing Events and Public Opinion: Evidence the Deepwater Horizon Disaster.” Political Behaviour 36: 1–22.

Blair, Benjamin D., Christopher M. Weible, Tanya Heikkila, and Larkin McCormack. 2016. “Certainty and Uncertainty in Framing the Risks and Benefits of Hydraulic Fracturing in the Colorado News Media.” Risk, Hazards & Crisis in Public Policy 6 (3): 290–307.

Blei, David M. 2012. “Probabilistic Topic Models.” Communications of the ACM 55, No. 4: 77– 84.

136

Bockstael, N. 1996. “Modeling Economics and Ecology: The Importance of a Spatial Perspective.” American Journal of Agricultural Economics 78: 1168–1180.

Boettcher, William A. 2004. “The Prospects for Prospect Theory: An Empirical Evaluation of International Relations Applications of Framing and Loss Aversion.” Political Psychology 25 (3): 331–362.

Bouvier, Yves. 2016. “Nuclear Fear in Europe: From Weapons to Power Stations.” Encyclopédie pour une histoire nouvelle de l'Europe. http://ehne.fr/en/article/material-civilization/risks- and-security/nuclear-fear-europe-weapons-power-stations.

Boykoff, Maxwell and Jules Boykoff. 2004. “Balance as Bias: Global Warming and the US Prestige Press. Global Environmental Change 14: 125–136.

Boyle, Michael P., Douglas M. McLeod, and Cory L. Armstrong. “Adherence to the Protest Paradigm: The Influence of Protest Goals and Tactics on News Coverage in U.S. and International Newspapers.” The International Journal of Press/Politics 17 (2): 127–144.

Brancati, Dawn. 2007. “Political Aftershocks: The Impact of Earthquakes on Intrastate Conflict.” The Journal of Conflict Resolution 51 (5): 715–743.

Breeze, Ruth. 2012. “Legitimation in Corporate Discourse: Oil Corporations after Deepwater Horizon.” Discourse & Society 23 (1): 3–18.

Brewer, Paul R. and Kimberly Gross. 2005. “Values, Framing, and Citizens' Thoughts about Policy Issues.” Political Psychology 26: 929–948.

Brewer, Paul R., Joseph Graf, and Lars Willnat. 2003. “Priming or Framing. Media Influence on Attitudes toward Foreign Countries.” Gazette: The International Journal for Communication Studies 65 (6): 493–508.

Brooks, Mark. 2014. “The Cautionary Tale of Kalamazoo.” Alternatives. http://www.alternativesjournal.ca/energy-and-resources/cautionary-tale-kalamazoo.

Brown, J. 2005. “To the Rescue.” Government Technology 18(2): 44–46.

Brown, Kate Pride. 2016. “The Prospectus of Activism: Discerning and Delimiting Imagined Possibility,” Social Movement Studies 15 (6): 547–560.

Brumfiel, Geoff. 2012. “Fukushima’s Uncertainty Problem.” Nature, 18 July. http://www.nature.com/news/fukushima-s-uncertainty-problem-1.11031#auth-1.

137

Brunken, Brigette Lynn. 2006. “Hurricane Katrina: A Content Analysis of Media Framing, Attribute Agenda Setting, and Tone of Government Response.” MA Thesis, Louisiana State University.

Bucher, Rue. 1957. “Blame and Hostility in Disaster.” American Journal of Sociology 62 (5): 467–475.

Buckingham, Susan. 2008. “The Rise of Environmental Politics and the Environmental Movement.” In Understanding Environmental Issues, edited by Susan Buckingham and Mike Turner. London: Sage.

Burgoon, Michael, Michael D. Miller, Marshall Cohen, and Charles L. Montgomery. 1978. “An Empirical Test of a Model of Resistance to Persuasion.” Human Communication Research 5 (1): 27–39.

Burton, I., and R. Kates. 1964. “The Perception of Natural Hazards in Resource Management.” Natural Resources Journal 3: 412–441.

Burton, I., R. Kates, and G. White. 1978. The Environment as Hazard. New York: Oxford University Press.

Busenberg, George. 2001. “Learning in Organizations and Public Policy.” Journal of Public Policy 21 (2): 173–189.

Button, Gregory. 2002. “Popular Media Reframing of Man-made Disasters.” In Catastrophe and Culture: The Anthropology of Disaster, edited by Susanna Hoffman and Anthony Oliver- Smith, 143–158. Santa Fe, NM: School for Advanced Research Press.

Cable, Sherry, and Thomas Shriver. 1995. “Production and Extrapolation of Meaning in the Environmental Justice Movement.” Sociological Spectrum15: 419–442.

Camerer, Colin and Martin Weber. 1992. “Recent Developments in Modeling Preferences: Uncertainty and Ambiguity.” Journal of Risk and Uncertainty 5 (1992): 325–370.

Carey, Sabine C. 2006. “The Dynamic Relationship between Protest and Repression.” Political Research Quarterly 59 (1): 1–11.

Carlin, Ryan, Gregory Love, and Elizabeth Zechmeister. 2014. “Natural Disaster and Democratic Legitimacy: The Public Opinion Consequences of Chile's 2010 Earthquake and Tsunami.” Political Research Quarterly 67 (1): 3–15.

138

Carroll, William and Robert Ratner. 1996. “Master Frames and Counter-hegemony: Political Sensibilities in contemporary social movements.” Canadian Review of Sociology 33 (4): 407–435.

Cart, Julie. 2010. “Gulf oil spill: Threat to Wildlife Grows.” Los Angeles Times, 29 April. http://latimesblogs.latimes.com/greenspace/2010/04/gulf-oil-spill-threat-to-wildlife- grows.html.

Carruthers, David and Patricia Rodriguez. 2009. “Mapuche Protest, Environmental Conflict and Social Movement Linkage in Chile.” Third World Quarterly 30 (4): 743–760.

Carvalho, Anabela. 2007. “Ideological Cultures and Media Discourses on Scientific Knowledge: Re-Reading News on Climate Change.” Public Understanding of Science 16 (2): 223– 243.

Catino, Maurizio. 2008. “A Review of Literature: Individual Blame vs. Organizational Function Logics in Accident Analysis.” Journal of Contingencies and Crisis Management 16 (1): 53–62.

Cederman, Lars-Erik, Nils B. Weidmann, and Kristian S. Gleditsch. 2011. “Horizontal Inequalities and Ethnonationalist Civil War: A Global Comparison.” American Political Science Review 105 (3): 478–495.

Chen, Ni. 2009. “Institutionalizing Public Relations: A Case Study of Chinese Government Crisis Communication on the 2008 Sichuan Earthquake.” Public Relations Review 35: 187–198.

Chenoweth, Erica and Kathleen Cunningham. 2013. “Understanding Nonviolent Resistance: An Introduction.” Journal of Peace Research 50 (3): 271–276.

Chenoweth, Erica and Maria Stephan. 2011. Why Civil Resistance Works: The Strategic Logic of Nonviolent Conflict. New York: Columbia University Press.

Chenoweth, Erica and Jay Ulfelder. 2017. “Can Structural Conditions Explain the Onset of Nonviolent Uprisings?” Journal of Conflict Resolution 61 (2): 298–324.

Cherry, Miriam and Judd Sneirson. 2011. “Beyond Profit: Rethinking Corporate Social Responsibility and Greenwashing after the BP Oil Disaster.” Tulane Law Review 85 (983): 983–1038.

Chomitz, Kenneth M., and David A. Gray. 1995. “Roads, Land Use, and Deforestation: A Spatial Model Applied to Belize.” The World Bank Economic Review 10: 487–512.

139

Chong, Dennis and James N. Druckman. 2007. “A Theory of Framing and Opinion Formation in Competitive Elite Environments.” Journal of Communication 57 (1): 99–118.

______. 2012. “Counterframing Effects.” The Journal of Politics 75 (1): 1–16.

Chong, Dennis. 1996. “Creating Common Frames of Reference on Political Issues.” In Political Persuasion and Attitude Change, edited by Diana C. Mutz, Paul M. Sniderman, and Richard A. Brody, 195–224. Ann Arbor: The University of Michigan Press.

Cho, Seung Ho and Karla K. Gower. 2006. “Framing Effect on the Public’s Response to Crisis: Human Interest Frame and Crisis Type Influencing Responsibility and Blame.” Public Relations Review 32: 420–422.

CBC. 2018. “Protesters rally across Canada over Trans Mountain pipeline decision.” CBC News. 4 June. https://www.cbc.ca/news/canada/british-columbia/protesters-rally-across-canada- over-trans-mountain-pipeline-decision-1.4691338.

Cleveland, Kyle. 2014. “The Fukushima Nuclear Crisis and the Politics of Uncertainty.” The Asia-Pacific Journal 12 (7). http://www.globalresearch.ca/the-fukushima-nuclear-crisis- and-the-politics-of-uncertainty-most-significant-public-health-crisis-in-modern- history/5369344.

Çoban, Aykut. 2004. “Community-based Ecological Resistance: The Bergama Movement in Turkey.” Environmental Politics 13 (2): 438–460.

Cohen, Charles and Eric Werker. 2008. “The Political Economy of “Natural” Disasters.” Harvard Business School Working Paper 08-040: 1–46.

Collins, A. M., and E. F. Loftus. 1975. “A Spreading-activation Theory of Semantic Processing.” Psychological Review 82: 407–428.

Comfort, Louise K. 2002. “Rethinking Security: Organizational Fragility in Extreme Events.” Public Administration Review 62: 98–107.

Comfort, Louise K., Yesim Sungu, David Johnson, and Mark Dunn. 2001. “Complex Systems in Crisis: Anticipation and Resilience in Dynamic Environments.” Journal of Contingencies and Crisis Management 9 (3): 144–158.

Cooper, Alice. 2002. “Media Framing and Social Movement Mobilization: German Peace Protest against INF Missiles, the Gulf War, and NATO Peace Enforcement in Bosnia.” European Journal of Political Research 41: 37–80.

140

Corrigall-Brown, Catherine. 2012. “The Power of Pictures: Images of Politics and Protest.” American Behavioral Scientist 56 (2): 131–134.

Corrigall-Brown, Catherine and Rima Wilkes. 2012. “Picturing Protest: The Visual Framing of Collective Action by First Nations in Canada.” American Behavioral Scientist 56 (2): 223–243.

______. 2014. “Media Exposure and the Engaged Citizen: How the Media Shape Political Participation.” The Social Science Journal 51: 408–421.

Craglia, Massimo, Robert Haining, and Paul Wiles. 2000. “A Comparative Evaluation of Approaches to Urban Crime Pattern Analysis.” Urban Studies 37 (4).

CRED. “The Emergency Events Database (EM-DAT).” http://www.emdat.be/.

Cutter, S. 2005. “Are We Asking the Right Question?” In What Is a Disaster: New Answers to Old Questions, edited by R.W. Perry and E.L. Quarantelli, 39–48. Philadelphia: Xlibris.

Curley, Shawn, J. Frank Yates, and Richard Abrams. 1986. “Psychological Sources of Ambiguity Avoidance.” Organizational Behavior and Human Decision Processes 38: 230–256.

Daley, Patrick and Dan O'Neill. 1991. “Sad Is Too Mild a Word: Press Coverage of the Exxon Valdez Oil Spill.” Journal of Communication 41 (4): 42–57.

Dalton, Russell J., Alix van Sickle, and Steven Weldon. 2009. “The Individual–Institutional Nexus of Protest Behaviour.” British Journal of Political Science 40: 51–73.

Dauncey, Guy. 2002. “2- Years in BC’s Environmental Movement.” Common Ground Magazine.

Dauvergne, Peter. 2017. “Is the Power of Brand-Focused Activism Rising? The Case of Tropical Deforestation.” Journal of Environment and Development 26 (2): 135–155.

Dauvergne, Peter and Justin Alger. 2017. “The Politics of Pacific Ocean Conservation: Lessons from the Pitcairn Islands Marine Reserve.” Pacific Affairs 90 (1): 29–50.

Dauvergne, Peter and Kate J. Neville. 2011. “Mindbombs of Right and Wrong: Cycles of Contention in the Activist Campaign to Stop Canada’s Seal Hunt.” Environmental Politics 20 (2): 192–209.

Davenport, Christian. 2007. “State Repression and Political Order.” Annual Review of Political Science 10: 1–23.

141

Dawson, Jane. 1996. Eco-Nationalism. Anti-nuclear Activism and National Identity in Russia, Lithuania, and Ukraine. Durham and London: Duke University Press.

Deacon, D., M. Pickering, P. Golding, and G. Murdock. 2007. Researching Communications. London: Hodder Education.

Del Sesto, Steven L. 1980. “Conflicting Ideologies of Nuclear Power: Congressional Testimony on Nuclear Power.” Public Policy 28 (1): 39–70. della Porta, Donatella and Mario Diani. 2006. Social Movements: An Introduction. Malden, MA: Blackwell Publishing.

De Marchi, B., S. Funtowicz, and J. Ravetz. 1996. “Seveso: A Paradoxical Classic Disaster.” In The Long Road to Recovery, edited by James Mitchell. Tokyo: United Nations University Press.

Dennis, Samuel F. Jr. 2006. “Prospects for Qualitative GIS at the Intersection of Youth.” Environment and Planning A 38: 2039‐2054.

Diani, Mario. 1995. Green Networks. A Structural Analysis of the Italian Environmental Movement. Edinburgh: Edinburgh University Press.

Dieckmann, Nathan, Robin Gregory, Ellen Peters, and Robert Hartman. “Seeing What You Want to See: How Imprecise Uncertainty Ranges Enhance Motivated Reasoning.” Risk Analysis 37 (3): 471–486.

Ditto, Peter H. and David F. Lopez. 1992. “Motivated Skepticism: Use of Differential Decision Criteria for Preferred and Nonpreferred Conclusions.” Journal of Personality and Social Psychology 63 (4): 568–584.

Dougherty, Michael. 2011. “The Global Gold Mining Industry, Junior Firms, and Civil Society Resistance in Guatemala.” Bulletin of Latin American Research 30 (4): 403–418.

Drabeck, Thomas and Enrico Quarantelli. 1967. “Scapegoats, Villains, and Disasters.” Trans- Action (March 1967): 12–17.

Drisco, James and Tina Maschi. 2016. Content Analysis. New York: Oxford University Press.

Druckman, James and Rose McDermott. 2008. “Emotion and the Framing of Risky Choice.” Political Behaviour 30: 297–321.

Drury, Cooper and Richard Stuart Olson. 1998. “Disasters and Political Unrest: An Empirical Investigation.” Journal of Contingencies and Crisis Management 6 (3): 153–161.

142

Duffy, David Cameron and Linda Elliott. 2010. “Oil and Hawaiian Seabirds.” Elepaio 70 (6): 41–43.

Dunford, Richard W. and Miranda L. Freeman. 2001. “A Statistical Model for Estimating Natural Resource Damages from Oil Spills.” International Oil Spill Conference Proceedings 2001 (1): 225–229.

Dunwoody, Sharon. 1999. “Scientists, Journalists, and the Meaning of Uncertainty.” In Communicating Uncertainty: Media Coverage of New and Controversial Science, edited by Sharon M. Friedman, Sharon Dunwoody, and Carol L. Rogers, 59–79. Mahwah, NJ: Lawrence Erlbaum Associates.

Dusa, Adrian. 2007. “User Manual for the QCA(GUI) Package in R.” Journal of Business Research 60 (5): 576–586.

Earthworks. 2012. U.S. Copper Porphyry Mines: The Track Record of Water Quality Impacts Resulting from Pipeline Spills, Tailings Failures and Water Collection and Treatment Failures. Earthworks.

Eckstein, Susan. 2001. Power and Popular Protest. Latin American Social Movements. Berkeley: University of California Press.

Edelman, Murray. 1985. “Political Language and Political Reality.” Policy Studies 18: 10–19.

Edelstein, Alex S. 1993. “Thinking about the Criterion Variable in Agenda-setting Research.” Journal of Communication 43: 85–99.

Elliott, David. 2013. Fukushima: Impacts and Implications. Basingstoke: Palgrave Macmillan.

EMC. 2010. “The State of Michigan. EMC Service, Support, and Storage Help Coordinate Urgent Response to Massive Oil Spill.” EMC. https://www.emc.com/collateral/customer- profiles/h7485-michigan-kalamazoo-cp.pdf.

Entman, Robert. 2003. “Cascading Activation: Contesting the White House’s Frame after 9/11.” Political Communication 20: 415–432.

______. 1993. “Framing: Toward Clarification of a Fractured Paradigm.” Journal of Communication 43, No. 4: 51–58.

EPA. “EPA Response to Kingston TVA Coal Ash Spill.” United States Environmental Protection Agency. https://www.epa.gov/tn/epa-response-kingston-tva-coal-ash-spill.

143

______. 1995. Human Health and Environmental Damages from Mining and Mineral Processing Wastes. Office of Solid Waste U.S. Environmental Protection Agency.

______. 1997. Damage Cases and Environmental Releases from Mines and Mineral Processing Sites, U.S. Environmental Protection Agency. Washington DC: EPA.

Erikson, Robert S. Michael B. MacKuen, and James A. Stimson. 2002. The Macro Polity. New York: Cambridge University Press.

European Commission. “Accident Report.” The Major Accident Reporting System. https://emars.jrc.ec.europa.eu/fileadmin/eMARS_Site/PhpPages/ViewAccident/ViewAcc identPublic.php?accident_code=664.

______. 2010. “Impacts of Gold Extraction in the EU.” European Commission – EUROSTAT, April 2. http://ec.europa.eu/environment/waste/mining/pdf/IH_2010-001.pdf.

European Environment Agency. 2009. “Tisza Spill Case Study.” Data and Maps. https://www.eea.europa.eu/data-and-maps/figures/spread-of-the-cyanide-spill-from-baia- mare/tisza_spill_case_study.eps.

European Environmental Agency. 2010. “Mapping the Impacts of Natural Hazards and Technological Accidents in Europe.” EEA Technical Report, No. 13/2010. www.eea.europa.eu/publications/mapping-the...of.../mapping-the-impacts-of-the.pdf.

Fan, Lilianne. 2013. “Disaster as Opportunity? Building Back Better in Aceh, Myanmar and Haiti.” HPG Working Paper. https://www.odi.org/sites/odi.org.uk/files/odi- assets/publications-opinion-files/8693.pdf.

Farrell, Thomas and G. Thomas Goodnight. 1981. “Accidental Rhetoric: The Root Metaphors of Three Mile Island.” Communication Monographs 48 (4): 271–300.

Fernandez, Susana Aguilar and Ana Ballesteros Peña. 2004. “Debating the Concept of Political Opportunities in Relation to the Galician Social Movement 'Nunca Máis'.” South European Society and Politics 9 (3): 28–53

Fillieule, Olivier. 2003. “France.” In Environmental Protest in Western Europe, edited by Christopher Rootes, 59–79. Oxford: Oxford University Press.

Fisher, William F. 1997. “Doing Good? The Politics and Anti-Politics of NGO Practices.” Annual Review of Anthropology 26: 439–464.

Fisheries Research Services. 2003. “Ten Years on – MV Braer Oil Spill.” The Government of Scotland. http://www.gov.scot/Uploads/Documents/AE17Braer.pdf.

144

Flores, Alejandro Quiroz and Alastair Smith. 2013. “Leader Survival and Natural Disasters.” British Journal of Political Science: 1–23.

Forbes, G.L. and D. M. Campbell. 1994. “The Braer Oil Spill Incident – Shetland, January 1993.” International Journal of Environmental Health Research 4 (1): 48–59.

Fortun, Kim. 2001. Advocacy after Bhopal: Environmentalism, Disaster, New Global Orders. Chicago: University of Chicago Press.

______. 2004. “From Bhopal to the Informating of Environmentalism: Risk Communication in Historical Perspective.” Osiris 19: 283–296.

Fox, Holly and Mark Hallam. 2011. “German Cabinet Approves 2022 Nuclear Shutdown.” Deutsche Welle, 7 June. http://www.dw.com/en/german-cabinet-approves-2022-nuclear- shutdown/a-15134028-1.

Fox, Craig and Martin Weber. 2002. “Ambiguity Aversion, Comparative Ignorance, and Decision Context.” Organizational Behavior and Human Decision Processes 88 (1): 476–498.

Fredriksson, Per, and Daniel Millimet. 2004. “Electoral Rules and Environmental Policy.” Economics Letters 84 (2): 237–44.

Freudenburg, William and Robert Gramling. 2011. Blowout in the Gulf. The BP Oil Spill Disaster and the Future of Energy in America. Cambridge: MIT Press.

Fritz, Oliver, and Peter Skerfving. 2005. “Monitoring and Analysis of Power Line Failures: An Examples of the Role of GIS.” In Geographic Information Systems in Business, edited by J. B. Pick, 301–323. Hershey, PA: Idea Group Publishing.

Gamson, William A. and David S. Meyer. 1996. “Framing Political Opportunity.” In Comparative Perspectives on Social Movements, edited by Doug McAdam, John D. McCarthy, and Mayer N. Zald, 2751–290. New York: Cambridge University Press.

Gamson, William A., and A. Modigliani. 1989. “Media Discourse and Public Opinion on Nuclear Power: A Constructionist Approach.” American Journal of Sociology 95 (1): 1– 37.

______. 1987. “The Changing Culture of Affirmative Action.” In Research in Political Sociology, edited by R. G. Braungart and M. M. Braungart. Greenwich, CT: JAI Press.

Garner, Robert. 2004. Animals, Politics, and Morality: 2nd Edition. Manchester: Manchester University Press.

145

Gephart, Robert P., Jr. 1984. “Making Sense of Organizationally Based Environmental Disasters.” Journal of Management 10 (2): 205–225.

______. 1993. “The Textual Approach: Risk and Blame in Disaster Sensemaking.” The Academy of Management Journal 36 (6): 1465–1514.

Giannakopoulos, Thanos. 2013. “Framing the Fukushima Daiichi Nuclear Accident: An International, Comparative Study of News Frames in Der Spiegel, the Japan Times and the Los Angeles Times.” MA Thesis, University of Leicester.

Gillespie, D. and R.W. Perry. 1974. “An Integrated Systems and Emergent Norm Approach to Mass Emergencies.” Mass Emergencies 1: 303–312.

Gimond, Manuel. 2018. “Intro to GIS and Spatial Analysis.” https://mgimond.github.io/Spatial/.

Gitlin, T. 1980. The Whole World is Watching: Mass Media in the Making & Unmaking of the New Left. Berkeley, CA: University of California Press.

Giuliano, Elise. 2011. Constructing Grievance. Ethnic Nationalism in Russia's Republics. Ithaca, NY: Cornell University Press.

Goffman, Erving. 1974. Frame Analysis: An Essay on the Organization of Experience. New York, NY: Harper & Row.

Goldstone, Jack A. 1991. Revolution and Rebellion in the Early Modern World. Berkeley: University of California Press.

Goodchild, Michale, Luc Anselin, Richard Appelbaum, and Barbara Herr Harthorn. 2000. “Toward Spatially Integrated Social Science.” International Regional Science Review 23 (2): 139–159.

Gorke, Alexander. 2011. “Fukushima in the Heart of Germany: Reflections on German News Coverage during the Crisis.” Wochi Kochi Magazine, October. http://www.wochikochi.jp/english/relayessay/2011/10/germany-fukushima.php.

Graf, William. 1990. “Fluvial Dynamics of Thorium-230 in the Church Rock Event, Puerco River, New Mexico.” Annals of the Association of American Geographers 80 (3): 327– 342.

Grant, Sonia. 2014. “The Kalamazoo River Spill: Pipelines, Politics, and Economies of Knowledge.” MA Thesis, University of Toronto.

146

Gray, Jeffrey. 1987. The Psychology of Fear and Stress. Cambridge: Cambridge University Press.

Greco, Veronica and Derek Roger. 2003. “Uncertainty, Stress, and Health.” Personality and Individual Differences 34: 1057–1068.

Green, Donald and Alan Gerber. 2012. Field Experiments: Design, Analysis and Interpretation. New York: W.W. Norton and Company.

Greenley, Douglas A., Richard G. Walsh, and Robert A. Young. 1981. “Option Value: Empirical Evidence from a Case Study of Recreation and Water Quality.” The Quarterly Journal of Economics 96 (4): 657–667.

Greenpeace. 2016. “The Fukushima Daiichi Accident Current Status and Marine Impacts.” Greenpeace, February. http://www.greenpeace.org/japan/Global/japan/pdf/Fukushima_Research_Briefing.pdf.

Grimmer, Justin and Brandon M. Stewart. 2013. “Text as Data: The Promise and Pitfalls of Automatic Content Analysis Methods for Political Texts.” Political Analysis 21: 1–31.

Gross, Kimberly. 2008. “Framing Persuasive Appeals: Episodic and Thematic Framing, Emotional Response, and Policy Opinion.” Political Psychology 29 (2): 169–192.

Grupe, Dan and Jack Nitschke. 2013. “Uncertainty and Anticipation in Anxiety: An Integrated Neurobiological and Psychological Perspective.” Nature Reviews Neuroscience 14 (7): 488–501.

Guo Yue, Peng Ru, Jun Su, and Laura Diaz Anadon. 2015. “Not in My Backyard, But Not Far Away from Me: Local Acceptance of Wind Power in China.” Energy 82: 722–733.

Gurney, Joan Neff and Kathleen Tierney. 1982. “Relative Deprivation and Social Movements: A Critical Look at Twenty Years of Theory and Research. The Sociological Quarterly 23 (1): 33–47.

Gurr, Ted. 1970. Why Men Rebel. Princeton, N.J.: Princeton University Press.

Haab, Timothy and Kenneth McConnell. 2002. Valuing Environmental and Natural Resources. The Econometrics of Non-Market Valuation. Cheltenham, UK: Edward Elgar.

Haas, Ingrid J. and Willima A. Cunningham. 2014. “The Uncertainty Paradox: Perceived Threat Moderates the Impact of Uncertainty on Political Tolerance.” Political Psychology 35 (2): 291–302.

147

Hackbarth, G., and B. Mennecke. 2005. “Strategic Positioning of Location Applications for Geo- business.” In Geographic Information Systems in Business, edited by J. B. Pick, 198–210. Hershey, PA: Idea Group Publishing.

Haigh, Michel M., Michael Pfau, Jamie Danesi, Robert Tallmon, Tracy Bunko, Shannon Nyberg, Bertha Thompson, Chance Babin, Sal Cardella, Michael Mink, and Brian Temple. 2006. “A Comparison of Embedded and Nonembedded Print Coverage of the U.S. Invasion and Occupation of Iraq.” The International Journal of Press/Politics 11 (2): 139–153.

Halliday, Josh. 2017. “Fracking protesters vow to 'put their lives on line' after scaling rig.” The Guardian. 22 October. https://www.theguardian.com/environment/2017/oct/22/fracking- protesters-vow-to-put-their-lives-on-line-after-scaling-rig.

Hanemann, W. Michael. 1994. “Valuing the Environment Through Contingent Valuation.” The Journal of Economic Perspectives 8 (4): 19–43.

Hanewitz, Cheryl. 2012. “Geographic Information Systems and the Political Process.” Western Political Science Association, 28 February. http://wpsa.research.pdx.edu/meet/2012/hanewicz.pdf.

Hannigan, John. 2012. Disasters without Borders. The International Politics of Natural Disasters. Cambridge: Polity.

Hartwig, Manuela, Sae Okura, Leslie Tkach-Kawasaki, and Yohei Kobashi. 2016. “Identifying the “Fukushima Effect”: Assessing Japanese Mass Media Coverage of International Nuclear Power Decisions.” Journal of International and Advanced Japanese Studies 8: 109–124.

Hasegawa, Koichi. 2014. “The Fukushima Nuclear Accident and Japan’s Civil Society: Context, Reactions, and Policy Impacts.” International Sociology 29 (4): 283–301.

Hausman, Jerry, Gregory K. Leonard, and Daniel McFadden. 1995. “A Utility-Consistent, Combined Discrete Choice and Count Data Model: Assessing Recreational Use Losses Due to Natural Resource Damage.” Journal of Public Economics 56 (1): 1–30.

Hellenic Shipping News. 2016. “Backgrounder: The Tragedy of Greek Tanker Irenes Serenade.” 14 November. http://www.hellenicshippingnews.com/backgounder-the-tragedy-of-greek- tanker-irenes-serenade/.

Hernan, Robert E. 2010. This Borrowed Earth: Lessons from the Fifteen Worst Environmental Disasters around the World. New York: Palgrave Macmillan.

148

Hill, Elizabeth M., Jessica Jenkins, and Lisa Farmer. 2008. “Family Unpredictability, Future Discounting, and Risk Taking.” The Journal of Socio-Economics 37: 1381–1396.

Hill, Elizabeth M., Lisa Thomson Ross, and Bobbi S. Low. 1997. “The Role of Future Unpredictability in Human Risk-Taking.” Human Nature 8 (4): 287–325.

Hoberg, George. 2013. “The Battle Over Oil Sands Access to Tidewater: A Political Risk Analysis of Pipeline Alternatives.” Canadian Public Policy 39 (3): 371–391.

Hoffman, Susanna. 2002. “The Monster and the Mother. The Symbolism of Disasters.” In Catastrophe and Culture: The Anthropology of Disaster, edited by Susanna Hoffman and Anthony Oliver-Smith, 113–141. Santa Fe, NM: School for Advanced Research Press.

Hoffman, Andrew and P. Devereaux Jennings. 2010. “The BP Oil Spill as a Cultural Anomaly? Institutional Context, Conflict and Change.” Ross School of Business Working Paper Working Paper No. 1151: 1–36.

Holba, Carrie. 2014. “Exxon Valdez Oil Spill.” Alaska Resources Library & Information Service. Updated 24 March. http://www.arlis.org/docs/vol2/a/EVOS_FAQs.pdf.

Homer-Dixon, Thomas. 1991. “On the Threshold: Environmental Changes as Causes of Acute Conflict.” International Security 16: 76–116.

______. 1994. “Environmental Scarcities and Violent Conflict.” International Security 19 (1): 5–40.

Horn, Stuart A. and Captain Phillip Neal. 1981. “The Atlantic Empress Sinking – A Large Spill without Environmental Disaster.” International Oil Spill Conference Proceedings 1981 (1): 429–435.

Horn, M. 1999. “GIS and the Geography of Politics.” In Geographical Information Systems, Volume 1, edited by Paul A. Longley, Michael F. Goodchild, David J. Maguire, and David W. Rhind, 939–951. New York: Wiley & Sons, Inc.

Hu, Connor and Dustin Tingley. 2015. “Who Are These People?? Evaluating the Demographic Characteristics and Political Preferences of Mturk Survey Respondents.” Research & Politics 2 (3).

Hulme, M. 2009. Why We Disagree about Climate Change: Understanding Controversy, Inaction and Opportunity. Cambridge: Cambridge University Press.

149

Hume, Mark. 2015. “Pollutants from Mount Polley breach may have long-term effects: study.” The Globe and Mail, 5 May. http://www.theglobeandmail.com/news/british- columbia/pollutants-from-mount-polley-breach-may-have-long-term-effects- study/article24272988/.

Hummel, Monte. 2016. “Environmental and Conservation Movements.” Historica Canada. https://www.thecanadianencyclopedia.ca/en/article/environmental-and-conservation- movements/.

Hunter, Justine and Mark Humer. 2015. “Design failure caused Mount Polley tailings breach, expert panel concludes.” The Globe and Mail, 30 January. http://www.theglobeandmail.com/news/british-columbia/design-failure-caused-mount- polley-tailings-breach-expert-panel-concludes/article22719967/.

IEM. 2010. A Study of the Economic Impact of the Deepwater Horizon Oil Spill. IEM.

Ilie, Luiza and Radu Sigheti. 2013. “Thousands rally in Romania against gold mine.” Reuters. 1 September. https://www.reuters.com/article/us-romania-goldmine/thousands-rally-in- romania-against-gold-mine-idUSBRE9800FU20130901.

Inglehart, Ronald, and Christian Welzel. 2005. Modernization, Cultural Change, and Democracy: The Human Development Sequence. Cambridge, UK: Cambridge University Press.

Ioannides, Y.M. 2000. “Economic Geography and the Spatial Evolution of Wages in the United States.” In Advances in Spatial Econometrics 1, edited by L. Anselin and R. Florax. Heidelberg: Springer-Verlag.

Iyengar, Shanto. 2010. “Framing Research: The Next Steps.” In Winning with Words: The Origins and Impact of Framing, edited by Brian F. Schaffner and Patrick J. Sellers. New York: Routledge.

______. 1990. “The Accessibility Bias in Politics: Television News and Public Opinion.” International Journal of Public Opinion Research 2 (1):1–15.

Iyengar, Shanto and D. R. Kinder. 1987. News that Matters: Television and American Opinion. Chicago, IL: University of Chicago Press.

Ivanov, A. Yu and V. V. Zatyagalova. 2008. “A GIS Approach to Mapping Oil Spills in a Marine Environment.” Journal International Journal of Remote Sensing 29 (21): 6297– 6313.

150

Jasper, James and Jane Poulsen. 1995. “Recruiting Strangers and Friends: Moral Shocks and Social Networks in Animal Rights and Anti-Nuclear Protests.” Soc. Problems 42: 493– 512.

Javeline, Debra. 2003. “The Role of Blame in Collective Action: Evidence from Russia.” American Political Science Review 97 (1): 107–121.

Jenks, R. H. and J. M. Malecki. 2004. “GIS - A Proven Tool for Public Health Analysis.” Journal of Environmental Health 67 (3): 32–34.

Jernelov, Arne. 2010. “The Threats from Oil Spills: Now, Then, and in the Future.” Ambio 39: 353–366.

Jernelov, Arne and Olof Linden. 1981. “Ixtoc I: A Case Study of the World's Largest Oil Spill.” Ambio 10 (6): 299–306.

Joint Committee on Atomic Energy. 1961. SL-1 Accident. Atomic Energy Commission Investigation Board Report. Washington, DC: U.S. Government Printing Office.

Jones, Eric C. and Arthur D. Murphy, eds. 2009. The Political Economy of Hazards and Disasters. Lanham, MD: AltaMira Press.

Jonsson, Patrik. 2010. “Worldwide BP Protest Day vilifies BP for Gulf oil spill.” Christian Science Monitor, 12 June. http://www.csmonitor.com/USA/2010/0612/Worldwide-BP- Protest-Day-vilifies-BP-for-Gulf-oil-spill.

Johnson, E., and A. Tversky. 1983. “Affect, Generalization, and the Perception of Risk.” Journal of Personality and Social Psychology 45: 20–31.

Johnson, Erik and John D. McCarthy. 2005. “The Sequencing of Transnational and National Social Movement Mobilization: The Organizational Mobilization of the Global and U.S. Environmental Movements.” In Transnational Protest and Global Activism, edited by Donatella Della Porta and Sideny Tarrow, 71–94. Oxford: Rowman and Littlefield.

Jost, John, Jack Glaser, Arie Kruglanski, and frank Sulloway. 2003. “Political Conservatism as Motivated Social Cognition.” Psychological Bulletin 129 (3): 339–375.

Joyce, Karen E., Stella E. Belliss, Sergey V. Samsonov, Stephen J. McNeill, and Phil J. Glassey. 2009. “A Review of the Status of Satellite Remote Sensing and Image Processing Techniques for Mapping Natural Hazards and Disasters.” Progress in Physical Geography 33(2): 183–207.

151

Kahan, Dan. 2013. “Ideology, Motivated Reasoning, and Cognitive Reflection.” Judgment and Decision Making 8 (4): 407–424.

Kahan, Dan, Hank Jenkins-Smith, and Donald Braman. 2011. “Cultural Cognition of Scientific Consensus.” Journal of Risk Research 14 (2): 147–174.

Kahneman, Daniel and Jack L. Knetsch. 1992. “Valuing Public Goods: The Purchase of Moral Satisfaction.” Journal of Environmental Economics and Management 22 (1): 57–70.

Kahneman, Daniel and Amos Tversky. 1979. “Prospect Theory: An Analysis of Decision under Risk.” Econometrica 47 (2): 263–291.

Kamieniecki, Sheldon. 1993. Environmental Politics in the International Arena: Movements, Parties, Organizations, and Policy. Albany, NY: State University of New York Press.

Kaplan, Robert. 1994. “The Coming Anarchy.” The Atlantic Monthly: 44–76.

Karlberg, Michael. 1997. “News and Conflict: How Adversarial News Frames Limit Public Understanding of Environmental Issues.” Alternatives Journal 23 (1): 22–27.

Keck, Margaret and Kathryn Sikkink. 1998. Activists beyond Borders: Advocacy Networks in International Politics. Ithaca, NY: Cornell University Press.

Kellow, Aynsley. 2000. “Norms, Interests and Environmental NGOs: The Limits of Cosmopolitanism.” Environmental Politics 9 (3): 1–22.

______. 1999. “Transnational Advocacy Networks in International and Regional Politics.” International Social Science Journal 51 (159): 89–101.

Kelman, Ilan. 2006. “Acting on Disaster Diplomacy.” Journal of International Affairs 59 (2): 215–240.

Keller, James. 2014. “Lack of Confidence in B.C. from Alaskan Fishing Industry since Mount Polley Breach.” The Globe and Mail, 21 August. http://www.theglobeandmail.com/news/british-columbia/lack-of-confidence-in-bc-from- alaskan-fishing-industry-since-mount-polley-breach/article20163310/.

Kerbo, Harold. 1982. “Movements of Crisis and Movements of Affluence: A Critique of Deprivation and Resource Mobilization Theories.” The Journal of Conflict Resolution 26 (4): 645–663.

152

Keren, Gideon. 2011. “On the Definition and Possible Underpinnings of Framing Effects: A Brief Review and a Critical Evaluation.” In Perspectives on Framing, edited by Gideon Keren, 3–34. New York and Hove: Psychology Press.

Kerr, Richard, Eli Kintisch, Erik Stokstad and Lauren Schenkman. 2010. “Will Deepwater Horizon Set a New Standard for Catastrophe?” Science 328 (5979): 674–675.

Kim, Hyo J. and Glen T. Cameron. 2011. “Emotions Matter in Crisis. The Role of Anger and Sadness in the Publics’ Response to Crisis News Framing and Corporate Crisis Response.” Communication Research 38(6):826–855.

Kim, Sei-Hill, Dietram A. Scheufele, and James Shanahan. 2002. “Think about it This Way: Attribute Agenda Setting Function of the Press and the Public's Evaluation of a Local Issue.” Journalism & Mass Communication Quarterly 79 (1): 7–25.

King, Dennis and Marisa Mazzotta. “Dollar-based Ecosystem Valuation Methods.” http://www.ecosystemvaluation.org/dollar_based.htm.

Kitschelt, Herbert P. 1986. “Political Opportunity Structures and Political Protest: Anti-Nuclear Movements in Four Democracies.” British Journal of Political Science 16 (1): 57–85.

Kitzinger, Jenny. 1999. “Researching Risk and the Media.” Health, Risk & Society 1 (1): 55–69.

Klandermans, Bert. 1984. “Mobilization and Participation: Social-psychological Expansions of Resource Mobilization Theory.” American Sociological Review 49: 583–600.

Klar, Samara, Joshua Robison, and James Druckman. 2013. “Political Dynamics of Framing.” In New Directions in Media and Politics, edited by T. Ridout, 173–192. New York, Routledge.

Klaus, Krista. 2010. “Memorial Day Protest of BP Held at Clearwater Station.” Tampa Bay Times, 31 May. http://www.tbo.com/news/florida/memorial-day-protest-of-bp-held-at- clearwater-station-38905.

Kleinsteuber, Hans J. and Barbara Thomass. “German Media Landscape.” In European Media Governance. National and Regional Dimensions, edited by Georgios Terzis, 111–124. University of Chicago Press, Chicago, 2007.

Kondratyev, Kirill, Alexei Grigoryev, and Costas Varotsos. 2002. Environmental Disasters: Anthropogenic and Natural. Chichester, UK: Springer-Praxis.

153

Kraft, Michael E. and Bruce B. Clafy. 1991. “Citizen Participation and the NIMBY Syndrome: Public Response to Radioactive Waste Disposal.” The Western Political Quarterly 44: 299–328.

Kreutz, Joakim. 2012. “From Tremors to Talks: Do Natural Disasters Produce Ripe Moments for Resolving Separatist Conflicts?” International Interactions 38: 482–502.

Krippendorff, K. 2013. Content Analysis: An Introduction to its Methodology. Thousand Oaks, CA: Sage, 2013.

Krishnamurthy, Parthasarathy, Patrick Carter, and Edward Blair. 2001. “Attribute Framing and Goal Framing Effects in Health Decisions.” Organizational Behavior and Human Decision Processes 85 (2): 382–399.

Krugman, P. 1991. “Increasing Returns and Economic Geography.” Journal of Political Economy 99: 438–499.

———. 1996. “Urban Concentration: The Role of Increasing Returns and Transport Costs.” International Regional Science Review 19: 5–30.

Kuehn Robert. 2000. “A Taxonomy of Environmental Justice.” Environmental Law Report 30: 10681–10703.

Kunda, Ziva. 1990. “The Case for Motivated Reasoning.” Psychological Bulletin 108 (3): 480– 498.

Kurtz, Hilda. 2003. “Scale Frames and Counter-scale Frames: Constructing the Problem of Environmental Injustice.” Political Geography 22: 887–916.

Kurzman, Charles. 1996. “Structural Opportunity and Perceived Opportunity in Social- Movement Theory: The Iranian Revolution of 1979.” American Sociological Review 61 (1): 153–170.

Lagadec, Patrick. 1982. Major Technological Risk. Oxford, UK: Pergamon Press.

Lakoff, George and Mark Johnson. 1980. Metaphors We Live by. Chicago, IL: University of Chicago Press.

Larabee, Ann. 2000. Decade of Disaster. Urbana and Chicago: University of Illinois Press.

Law, Robin J. and Carole Kelly. 2004. “The Impact of the “Sea Empress” Oil Spill.” Aquatic Living Resources 17: 389–394.

154

Le Billon, Philippe and Arno Waizenegger. 2007. “Peace in the Wake of Disaster: Secessionist Conflicts and the 2004 Indian Ocean Tsunami.” Transactions of the Institute of British Geographers 32 (3): 411–427.

Lee, Adrian. 2014. “Cleaning up after Canada’s Largest Tailings Pond Leak.” Maclean’s, 14 August. http://www.macleans.ca/news/canada/cleaning-up-after-canadas-largest-tailings- pond-leak/.

Leiserowitz, Anthony. 2005. “American Risk Perceptions: Is Climate Change Dangerous?” Risk Analysis 25 (6): 1433–1442.

Lendon, Hannah and Brian Martin. 2007. “Environmental Disasters.” In Justice Ignited: The Dynamics of Backfire. Plymouth UK: Rowman and Littlefield Publishing.

Leon, Carmelo J., Jorge E. Araña, W. M. Hanemann, and Pere Riera. 2014. “Heterogeneity and Emotions in the Valuation of Non-use Damages Caused by Oil Spills.” Ecological Economics 97: 129–139.

Leonard, Liam. 2008. The Environmental Movement in Ireland. Springer.

Lerner, Jennifer, Roxana Gonzalez, Deborah Small, and Baruch Fischhoff. 2003. “Effects of Fear and Anger on Perceived Risk of Terrorism: A National Field Experiment.” Psychological Science 14: 144–150.

Lerner, Jennifer and D. Keltner. 2000. “Beyond Valence: Toward a Model of Emotion-specific Influences on Judgment and Choice.” Cognition and Emotion, 14 (4): 473–493.

———.2001. “Fear, Anger, and Risk.” Journal of Personality and Social Psychology 81: 146– 159.

Levin, David. 2005. “Framing Peace Policies: The Competition for Resonant Themes.” Political Communication 22 (1): 83–108.

Levin, Irwin, Gary Gaeth, J. Schreiber, and M. Lauriola. 2002. “A New Look at Framing Effects: Distribution of Effect Sizes, Individual Differences, and Independence of Types of Effects.” Organizational Behavior and Human Decision Processes 88 (2): 411–429.

Levin, Irwin and Gary Gaeth. 1988. “How Consumers Are Affected by the Framing of Attribute Information before and after Consuming the Product.” Journal of Consumer Research 15 (3): 374–378.

Levin, Irwin P., Sandra L. Schneider, and Gary J. Gaeth. 1998. “All Frames Are Not Created Equal.” Organizational Behavior and Human Decision Processes 76: 149–188.

155

Levy, Jack. 1997. “Prospect Theory, Rational Choice, and International Relations.” International Studies Quarterly 41 (1): 87–112.

Liberatore, Angela. 1999. “The Management of Uncertainty Response and Learning Processes following Chernobyl.” Thesis, European University Institute.

Li, Xiaobing and Michael Molin, eds. 2014. Oil: A Cultural and Geographic Encyclopedia of Black Gold, Volume 1. Santa Barbara, Denver, Oxford: ABC-CLIO.

Lim, Jeongsub and Hyunjin Seo. 2008. “Frame Flow between Government and the News Media and its Effects on the Public: Framing of North Korea.” International Journal of Public Opinion Research 21 (2): 204–223.

Linnitt. Carol. 2015. “No Fines, No Charges Laid for Mount Polley Mine Disaster.” Desmog Canada, 18 December. http://www.desmog.ca/2015/12/18/no-fines-no-charges-laid- mount-polley-mine-disaster.

Lipset, Seymour. 1959. “Some Social Requisites of Democracy.” American Political Science Review 53 (1): 69–105.

Stephan J. Lynton. 1979. “Antinuclear protesters call for construction ban,” The Washington Post, 9 April: A13.

Loureiro, Maria L. and John B. Loomis. 2013. “International Public Preferences and Provision of Public Goods: Assessment of Passive Use Values in Large Oil Spills.” Environmental and Resource Economics 56 (4): 521–534.

Loureiro, Maria L., John B. Loomis, and Maria Xosé Vázquez. 2009. “Economic Valuation of Environmental Damages due to the Prestige Oil Spill in Spain.” Environmental Resource Economics 44: 537–553.

Loutit, J. F., W.G. Marley, and R.S. Russell. 1960. The Nuclear Reactor Accident at Windscale - October, 1957: Environmental Aspects. Westminster, London: Medical Research Council.

Lovan, Dylan. 2010. “After Decade, Still Signs of Coal Slurry spill.” Washington Post, 17 October. http://www.washingtonpost.com/wp- yn/content/article/2010/10/15/AR2010101507010.html.

Luke, Timothy. 1987. “Chernobyl: The Packaging of Transnational Ecological Disaster.” Critical Studies in Mass Communication 4 (4): 351–375.

156

Macdonald, William. 1980. “Ixtoc I: International and Domestic Remedies for Transboundary Pollution Injury.” Fordham Law Review 49 (3): 404–431.

Madrid, J. A. J., A. Garcia-Olivaresa, J. Ballabrera Poya, and E. Garcia-Ladonaa. “Managing Large Oil Spills in the Mediterranean.” Sparrho. https://www.sparrho.com/item/managing-large-oil-spills-in-the-mediterranean/8e8a9b/.

Maantay, Juliana and John Ziegler. 2006. GIS for the Urban Environment. Redlands: ESRI Press.

Maheswaran, Durairaj and Joan Meyers-Levy. 1990. “The Influence of Message Framing and Issue Involvement.” Journal of Marketing Research 27 (3): 361–367.

Marcus, G. E., R. Neuman, and M. MacKuen. 2000. Affective Intelligence and Political Judgment. Chicago: University of Chicago Press.

Mariani, Simone, Maria Elena Cefalì, Marc Terradas, Eglantine Chappuis, and Enric Ballesteros. 2014. “Using Catenas for GIS-Based Mapping of NW Mediterranean Littoral Habitats.” Estuarine, Coastal and Shelf Science 147: 56–67.

Martin, Lawrence. 1979. “U.S. rally turns anti-nuclear protest into national movement.” The Globe and Mail, 8 May.

Marris, Peter. 2005. The Politics of Uncertainty. London and New York: Routledge.

Marshall, Bob. “Oysters Are Uniquely Sensitive to Gulf of Mexico Oil Spill.” Times-Picayune, 25 May 2010. http://www.nola.com/news/gulf-oil- spill/index.ssf/2010/05/oysters_are_uniquely_sensitive.html.

Martin, Brian. Justice Ignited: The Dynamics of Backfire. Plymouth UK: Rowman and Littlefield Publishing, 2007.

Martin, Arthur. “Locked out: Oil Spill Victims Try to Shame BP into Paying Compensation to Fishermen One Year on from Deepwater Disaster.” Daily Mail, 15 April 2011. http://www.dailymail.co.uk/news/article-1376844/BP-oil-spill-protest-Victims-seeking- compensation-Gulf-Mexico-disaster-locked-out.html.

Martinez-Alier, Joan. “Global Environmental Justice and the Environmentalism of the Poor.” In the Oxford Handbook of Environmental Political Theory, edited by Teena Gabrielson, Cheryl Hall, John M. Meyer, and David Schlosberg. Oxford University Press, 2016.

157

Mas, Jean Francois, H. Pug, Jose Luis Palacio, and Atahualpa Sosa-Lopez. 2004. “Modelling Deforestation Using GIS and Artificial Neural Networks.” Environmental Modelling & Software 19 (5): 461–471.

Matthew, Richard. 2002. “In Defense of Environment and Security Research.” ECSP Report (8): 109–124.

McAdam, Doug, John McCarthy, and Mayer Zald. 1996. “Introduction.” In Comparative Perspectives on Social Movements: Political Opportunities, Mobilizing Structure, and Cultural Framings, edited by Doug McAdam, John McCarthy, and Mayer Zald. New York: Cambridge University Press.

McAdam, Doug, Sidney Tarrow and Charles Tilly. 2001. Dynamics of Contention. Cambridge: Cambridge University Press.

McCarthy, John and Mayer Zald. 1973. The Trend of Social Movements in America: Professionalization and Resource Mobilization. Morristown, NJ: General Learning Corporation.

______. 1977. “Resource Mobilization and Social Movements: A Partial Theory.” American Journal of Sociology 82 (6): 1212–1241.

McCay, Deborah French, Jill Jennings Rowe, Nicole Whittier, Sankar Sankaranarayanan, and Dagmar Schmidt Etkin. 2004. “Estimation of Potential Impacts and Natural Resource Damages of Oil.” Journal of Hazardous Materials 107: 11–25.

McClure, J., J. White, and CG. Sibley. 2009. “Framing Effects on Preparation Intentions: Distinguishing Actions and Outcomes.” Disaster Prevention and Management 18: 187– 199.

McCombs, Maxwell E. and Donald L. Shaw. 1972. “The Agenda-setting Function of Mass Media.” Public Opinion Quarterly 36 (2): 176–187.

McCormick, John. 1991. Reclaiming Paradise. Bloomington and Indianapolis: Indiana University Press.

McGraw, Kathleen M. 1991. “Managing Blame: An Experimental Test of the Effects of Political Accounts.” American Political Science Review 85: 1133–1157.

McKelvie, Alexander, J. Michael Haynie, and Veronica Gustavsson. 2011. “Unpacking the Uncertainty Construct: Implications for Entrepreneurial Action.” Journal of Business Venturing 26: 273–292.

158

Macnamara J. 2005. “Media Content Analysis: Its Uses, Benefits and Best Practice Methodology.” Asia Pacific Public Relations Journal 6 (1): 1–34.

Medzihorsky, Juraj, Ioana-Elena Oana, Mario Quaranta and Carsten Q. Schneider. 2016. “SetMethods: Functions for Set-Theoretic Multi-Method Research and Advanced QCA.” R package version 2.1.

Mehta, Michael D. 2004. Risky Business: Nuclear Power and Public Protest in Canada. Lexington Books.

Meissner, Dirk. 2016. “Mount Polley Mine Disaster Hits 2-year Mark, Fallout Still Causes Divisions.” CBC News, 4 August. http://www.cbc.ca/news/canada/british- columbia/mount-polley-anniversary-1.3706850.

Mercer, Jonathan. 2005. “Prospect Theory and Political Science.” Annual Review of Political Science 8:1–21.

Metaxa-Kakavouli, Danae, Paige Maas, and Daniel P. Aldrich. “How Social Ties Influence Hurricane Evacuation Behavior.” Proceedings of the ACM on Human-Computer Interaction – CSCW 2 (CSCW), Article No. 122.

Mintz, Alex and Steven Redd. 2003. “Framing Effects in International Relations.” Synthese 135 (2): 193–213.

Mitchell, James. 1996. “Improving Community Responses to Industrial Disasters.” In The Long Road to Recovery, edited by James Mitchell. Tokyo: United Nations University Press.

Mitchell, Andy. 1999. The ESRI Guide to GIS Analysis. Volume 1: Geographic Patterns and Relationships. Redlands: Environmental Systems Research Institute, Inc.

Moldan, A., L. Jackson, S. McGibbon and J. Vander Westhuizen. 1985. “Some Aspects of the Castillo de Bellver Oilspill.” Marine Pollution Bulletin 16. (3): 97–102.

Molotch, Harvey. 1970. “Oil in Santa Barbara and Power in America.” Sociological Inquiry 40: 131–144.

Molotch, Harvey and Marilyn Lester. 1975. “Accidental News: The Great Oil Spill as Local Occurrence and National Event.” American Journal of Sociology 81 (2): 235–260.

Monbiot, George. 2011. “The True Value of Nature is Not a Number with a Pound Sign in Front.” The Guardian, 6 June. https://www.theguardian.com/commentisfree/2011/jun/06/monetisation-natural-world- definitive-neoliberal-triumph.

159

Montero, P., J. Blanco, J. M. Cabanas, J. Maneiro, Y.Pazos, and A. Morono. “Oil Spill Monitoring and Forecasting on the Prestige-Nassau Accident.” In Proceedings of the 26th Arctic and Marine Oil Spill Program (AMOP) Technical Seminar, 1013–1030. Ottawa: Environment, 2003.

Morenoff, Jeffrey, and Robert J. Sampson. 1997. “Violent Crime and the Spatial Dynamics of Neighborhood Transition: Chicago 1970-1990.” Social Forces 76: 31–64.

Morrison, Denton. 1971.“Some Notes toward Theory on Relative Deprivation, Social Movements, and Social Change.” The American Behavioral Scientist 14 (5): 675–690.

Morton, Thomas A., Anna Rabinovich, Dan Marshall, and Pamela Bretschneider. 2011. “The Future that May (or May Not) Come: How Framing Changes Responses to Uncertainty in Climate Change Communications.” Global Environmental Change 21: 103–109.

Mudder, T. I. and M. Botz. 2004. “Cyanide and Society: A Critical Review.” The European Journal of Mineral Processing and Environmental Protection 4 (1): 62–74.

Mullinix, Kevin J., Thomas Leeper, James Druckman, and Jeremy Freese. 2015. “The Generalizability of Survey Experiments.” Journal of Experimental Political Science 2 (2): 109–138.

Munilla, Ignacio, Jose M. Arcos, Daniel Oro, David Alvarez, Patricia Leyenda, and Alberto Velando. 2011. “Mass Mortality of Seabirds in the Aftermath of the Prestige Oil Spill.” Ecosphere 2 (7): 1–14.

Mutz, Diana. 2011. Population-Based Survey Experiments. Princeton: Princeton University Press.

Myer, Peter G. 1984. “IXTOC I: Case Study of a Major Oil Spill.” University of Rhode Island Theses and Major Papers (133): 1–39.

National Research Council. 2005. Valuing Ecosystem Services: Toward Better Environmental Decision-Making. The National Academies Press.

Nave, J. Gil. 2000. “The Politics of Environmental Groups in Portugal.” Thesis, European University Institute.

Neal, David. 1984. “Blame Assignment in a Diffuse Disaster Situation: A Case Example of the Role of an Emergent Citizen Group.” International Journal of Mass Emergencies and Disasters 2 (2): 251–266.

160

Nebehay, Stephanie. 2011. “Japan requests foreign rescue teams, UN says.” Reuters World News, 11 March. http://www.reuters.com/article/us-japan-quake-aid-refile- idUSTRE72A71320110311.

Nel, Philip and Marjolein Righarts. 2008. “Natural Disasters and the Risk of Violent Civil Conflict.” International Studies Quarterly 52: 159–185.

Nelson, Michael P., Jeremy T. Bruskotter, and John A. Vucetich. 2015. “Does nature have value beyond what it provides humans?” The Conversation, 2 October. https://theconversation.com/does-nature-have-value-beyond-what-it-provides-humans- 47825.

Nelson, N., K. P. Kitchen, and R.H. Maryon. “A Study of the Movement of Radioactive Material Discharged during the Windscale Fire in October 1957.” Atmospheric Environment 40 (2006): 58–75.

Nelson, Gerald C., and Daniel Hellerstein. 1997. “Do Roads Cause Deforestation? Using Satellite Images in Econometric Analysis of Land Use.” American Journal of Agricultural Economics 79: 80–88.

Neuman, W. Russell, Marion R. Just, and Ann N. Crigler. 1992. Common Knowledge. News and the Construction of Political Meaning. Chicago: The University of Chicago Press.

Neville, Kate J., and Erika Weinthal. 2016. “Scaling up Site Disputes: Strategies to Redefine ‘Local’ in the Fight against Fracking.” Environmental Politics 25 (4): 569–592.

Niggemeier, Jan. 2015. “Sayōnara Nukes - Development, Mobilisation Strategies and Organisation of Anti-Nuclear Protest in Japan after the 2011 Fukushima Accident.” Centre for East and South-East Asian Studies, Lund University, Sweden. Working Paper (48): 1–48.

NOAA. “NOAA Studies Documenting the Impacts of the Deepwater Horizon Oil Spill.” US National Oceanic and Atmospheric Administration, Office of Response and Restoration. http://response.restoration.noaa.gov/deepwater-horizon-oil-spill/noaa-studies- documenting-impacts-deepwater-horizon-oil-spill.html.

Noah, T. 1994. “Russian Oil Spill May Approach the Size of Exxon Valdez.” Wall Street Journal, 26 October: A22.

Norris, Pipa. 2002. Democratic Phoenix. Reinventing Political Activism. Cambridge: Cambridge University Press.

161

Norris, Pippa. 1995. “The Restless Searchlight: Network News Framing in the Post-Cold War World.” Political Communication 12 (4): 357–370.

Norse, Elliott and John Amos. 2010. “Impacts, Perception, and Policy Implications of the Deepwater Horizon Oil and Gas Disaster.” Environmental Law Reporter.

NRDC. 2015. “Summary of Information Concerning the Ecological and Economic Impacts of the BP Deepwater Horizon Oil Spill Disaster.” NRDC Issue Paper. https://www.nrdc.org/sites/default/files/gulfspill-impacts-summary-IP.pdf.

Nuclear Heritage. “Talvivaara mine: environmental disaster in Finland.” http://www.nuclear- heritage.net/index.php/Talvivaara_mine:_environmental_disaster_in_Finland#cite_note- 1.

Nuclear Safety Analysis Center. 1980. Analysis of Three Mile Island - Unit 2 Accident. Electric Power Research Institute.

OECD. 2018. “Income inequality.” https://data.oecd.org/inequality/income-inequality.htm.

OECD. 2018b. “Infant mortality rates.” https://data.oecd.org/healthstat/infant-mortality- rates.htm.

Oliver-Smith, Anthony. 1996. “Anthropological Research on Hazards and Disasters.” Annual Review of Anthropology 25: 303–328.

______. 2002. “Theorizing Disasters. Nature, Power, and Culture.” In Catastrophe and Culture: The Anthropology of Disaster, edited by Susanna Hoffman and Anthony Oliver-Smith, 23–48. Santa Fe, NM: School for Advanced Research Press.

Oliver-Smith, Anthony and Susanna M. Hoffman. 2002. “Introduction: Why Anthropologists Should Study Disasters.” In Catastrophe and Culture: The Anthropology of Disaster, edited by Susanna Hoffman and Anthony Oliver-Smith, 3–22. Santa Fe, NM: School for Advanced Research Press.

Olson, Richard Stuart. 2008. “Toward a Politics of Disaster: Losses, Values, Agendas, and Blame.” In Crisis Management, Volume III, edited by Arjen Boin. London: Sage.

O’Neill, Kate, Erika Weinthal, and Patrick Hunnicutt. 2017. “Seeing Complexity: Visualization Tools in Global Environmental Politics and Governance.” Journal of Environmental Studies and Sciences 7 (4): 490–506.

Opp, Karl-Dieter. 2009. Theories of Political Protest and Social Movements. New York: Routledge, 2009.

162

Opp, Karl-Dieter and Bernhard Kittel. 2010. “The Dynamics of Political Protest: Feedback Effects and Interdependence in the Explanation of Protest Participation.” European Sociological Review 26 (1): 97–109.

Pain, D. J., A. Sanchez, and A. A. Mehargc. 1998. “The Doñana Ecological Disaster: Contamination of a World Heritage Estuarine Marsh Ecosystem with Acidified Pyrite Mine Waste.” The Science of the Total Environment 222: 45–54.

Paine, Robert. 2002. “Danger and the No-Risk Thesis.” In Catastrophe and Culture: The Anthropology of Disaster, edited by Susanna Hoffman and Anthony Oliver-Smith, 67– 90. Santa Fe, NM: School for Advanced Research Press.

Pantti, Mervi Katriina and Karin Wahl-Jorgensen. 2011. “‘Not an Act of God’: Anger and Citizenship in Press Coverage of British Man-made Disasters.” Media, Culture & Society 33 (1): 105–122.

Pastor, Manuel, Jennifer Ito, and Rachel Rosner. 2011. Transactions, Transformations, Translations. Metrics That Matter for Building, Scaling, and Funding Social Movements. Los Angeles: USC Program for Environmental and Regional Equity.

Pastor, Dolors, Jaume Sanchez, Cinta Porte, and Joan Albaiges. 2001. “The Aegean Sea Oil Spill in the Galicia Coast (NW Spain). I. Distribution and Fate of the Crude Oil and Combustion Products in Subtidal Sediments.” Marine Pollution Bulletin 42 (10): 895– 904.

Pelling, Mark and Kathleen Dill. 2008. “Disaster Politics: From Social Control to Human Security.” Environment, Politics and Development Working Paper Series: 1–24.

______. 2010. “Disaster Politics: Tipping Points for Change in the Adaptation of Sociopolitical Regimes.” Progress in Human Geography 34 (1): 21–37.

Perez, Garcia. 2003. “Early Socio-political and Environmental Consequences of the Prestige Oil Spill in Galicia.” Disasters 27 (3): 207–223.

Perla, Hector, Jr. 2011. “Explaining Public Support for the Use of Military Force: The Impact of Reference Point Framing and Prospective Decision Making.” International Organization 65 (1): 139–167.

Perrow, Charles. 1979. “The Sixties Observed.” In The Dynamics of Social Movements, edited by Mayer Zald and John McCarthy. Cambridge, MA: Winthrop.

______. 1984. Normal Accidents: Living with High Risk Technologies. New York: Basic Books.

163

Petryna, Adriana. 1995. “Sarcophagus: Chernobyl in Historical Light.” Cultural Anthropology 10 (2): 196–220.

Pfau, M., M. Haigh, M. Gettle, M. Donnelly, G. Scott, and D. Warr. 2004. “Embedding Journalists in Military Combat Units: Impact on Newspaper Story Frames and Tone.” Journalism & Mass Communication Quarterly 81: 74–88.

Pinto, S. A Survey of the Underground Siting of Nuclear Power Plants. 1979. Wurenlingen: Swiss Federal Institute for Reactor Research.

Piven, Frances and Richard Cloward. 1977 Poor People's Movements: Why They Succeed, How They Fail. New York: Vintage Books.

Pop-Eleches, Grigore and Graeme B. Robertson. 2015. “Structural Conditions and Democratization.” Journal of Democracy 26 (3).

Pope, Carl. 2010. “Drill, Baby, Drill — Or Whoops, Baby, Spill?” The Huffington Post, 30 June. http://www.huffingtonpost.com/carl-pope/drill-baby-drill----or-wh_b_558473.html.

Prater, Carla and Michael Lindell. 2000. “Politics of Hazard Mitigation.” Natural Hazards Review 1 (2): 73–82.

Price, Vincent and David Tewksbury. 1997. “News Values and Public Opinion.” In Progress in Communication Sciences, Vol. 13, edited by George A. Barnett, and Franklin J. Boster. Greenwich, CT: Ablex Publishing Corporation.

Quarantelli, Enrico L. 1987. “Disaster Studies: An Analysis of the Social Historical Factors Affecting the Development of Research in the Area.” International Journal of Mass Emergencies and Disasters 5 (3): 285–310.

______. 1987. “Presidential Address: What Should We Study? Questions and Suggestions for Researchers about the Concept of Disasters.” International Journal of Mass Emergencies and Disasters 5: 7–32.

______. 2000. “Disaster Research.” In Encyclopedia of Sociology, edited by E. Borgatta and R. Montgomery, 682–688. New York: Macmillan.

______. 2005. “A Social Science Research Agenda for the Disasters of the 21st Century.” In What is a Disaster? New Answers to Old Questions, edited by R.W. Perry and E.L. Quarantelli, 325–396. Philadelphia: Xlibris.

Rabe, B. G. 1994. Beyond NIMBY: Hazardous Siting in Canada and the United States. Washington, DC: The Brookings Institute.

164

Rambaldi, Giacomo, Robert Chambers, Mike McCall, and Jefferson Fox. 2006. “Practical Ethics for PGIS Practitioners, Facilitators, Technology Intermediaries and Researchers.” Participatory Learning and Action 54: 106–113.

Randall, Alan. 1987. “Total Economic Value as a Basis for Policy.” Transactions of the American Fisheries Society 116: 325–335.

Rasmussen, Thomas H. 1992. “Not in My Backyard: The Politics of Siting Prisons, Landfills, and Incinerators.” State & Local Government Review 24 (3): 128–134.

Reed, Jennifer Bond. 2002. Love Canal. Philadelphia, PA: Chelsea House Publishers.

Richardson, Leslie and John Loomis. 2009. “The Total Economic Value of Threatened, Endangered and Rare Species: An Updated Meta-analysis.” Ecological Economics 68:1535–1548.

Richmond, Jeffrey. 2014. “Protesters Slam Imperial Metals over Mount Polley.” Mining.com, 12 August. http://www.mining.com/protesters-slam-imperial-metals-over-mount-polley/.

Riffe, Daniel, Stephen Lacy, and Frederick Fico. 1998. Analyzing Media Messages: Using Quantitative Content Analysis in Research. Mahwah, N.J. and London: Lawrence Erlbaum Associates.

Robinson, Scott E., Warren S. Eller, Melanie Gall, and Brian J. Gerber. 2013. “The Core and Periphery of Emergency Management Networks.” Public Management Review 15 (3): 344–362.

Rodriguez, Havidan, Enrico L. Quarantelli, and Russell Dynes. 2007. Handbook of Disaster Research. New York: Springer.

Rootes, Christopher. 2003. Environmental Protest in Western Europe. New York: Oxford University Press.

______. 2005. “A Limited Transnationalization? The British Environmental Movement.” In Transnational Protest and Global Activism, edited by Donatella Della Porta and Sideny Tarrow, 21–44. Oxford: Rowman and Littlefield.

Rose, Thomas and Trevor Sweeting. 2016. “How Safe is Nuclear Power? A Statistical Study Suggests Less Than Expected.” Bulletin of the Atomic Scientists 72 (2): 112–115.

Roser, Max and Esteban Ortiz-Ospina. 2016. “Income Inequality.” OurWorldInData. https://ourworldindata.org/income-inequality.

165

Rucht, Dieter. 1996. “The Impact of Environmental Movements in Western Societies.” In How Social Movements Matter, edited by Marco Giugni, Doug McAdam, and Charles Tilly. Minneapolis: University of Minnesota Press.

Ruiter, Robert, Charles Abraham and Gerjo Kok. 2001.“Scary Warnings and Rational Precautions: A Review of the Psychology of Fear Appeals.” Psychology & Health 16 (6): 613–630.

Rushe, Dominic. 2015. “BP Set to Pay Largest Environmental Fine in US History for Gulf Oil Spill.” The Guardian, 2 July. https://www.theguardian.com/environment/2015/jul/02/bp- will-pay-largest-environmental-fine-in-us-history-for-gulf-oil-spill.

Rweyendela, Amani. 2014. “The Communication of Social and Environmental Performance to Discharge Stakeholder Accountability: A Case Study of Talvivaara Mining Company Limited.” Master’s Thesis, University of Jyvaskyla School of Business and Economics.

Salehyan, Idean. 2008. “From Climate Change to Conflict? No Consensus Yet.” Journal of Peace Research 45 (3): 315–326.

Sampson, Robert J., Jeffrey Morenoff, and Felton Earls. 2000. “Beyond Social Capital: Neighborhood Mechanisms and Structural Sources of Collective Efficacy for Children.” American Sociological Review.

Saunders, Clare. 2013. Environmental Networks and Social Movement Theory. London: Bloomsbury Academic.

Schencking, J. Charles. 2008. “The Great Kanto Earthquake and the Culture of Catastrophe and Reconstruction in 1920s Japan.” The Journal of Japanese Studies 34 (2): 295–331.

Scheufele, Bertram T. 2004. “Framing-effects Approach: A Theoretical and Methodological Critique.” Communications 29: 401–428.

Scheufele, Dietram A. and Shanto Iyengar. 2014. “The State of Framing Research: A Call for New Directions.” In The Oxford Handbook of Political Communication Theories, edited by Kate Kenski and Kathleen Hall Jamieson. New York: Oxford University Press.

Scheufele, Dietram A., and D. Tewksbury. 2007. “Framing, Agenda Setting, and Priming: The Evolution of Three Media Effects Models.” Journal of Communication 57 (1): 9–20.

Schleifstein, Mark. 2016. “BP Oil Spill Cost Fishing Industry at Least $94.7 Million in 2010.” The Times-Picayune, 27 June. http://www.nola.com/environment/index.ssf/2016/06/bp_spill_cost_gulf_fishing_ind.htm l.

166

Schlosberg, David. 2004. “Reconceiving Environmental Justice: Global Movements and Political Theories.” Environmental Politics13 (3): 517–540.

Schussman, Alan and Sarah Soule. 2005. “Process and Protest: Accounting for Individual Protest Participation.” Social Forces 84 (2): 1083–1108.

Scott, Shaunna L., Stephanie McSpirit, Sharon Hardesty, and Robert Welch. 2005. “Post Disaster Interviews with Martin County Citizens: "Gray Clouds" of Blame and Distrust.” Journal of Appalachian Studies 11, No. 1/2 (Spring/Fall 2005): 7–29.

Schmidt, Jeremy. 2014. “Framing Environmental Crises: Correlating Action to Outcomes for the 1969 Santa Barbara and 2010 Deepwater Horizon Oil Spills.” Master’s Thesis, University of Northern British Columbia.

Sealey, Geraldine. 2000. “Sludge Spill Pollutes Ky., W. Va. Waters.” ABC News, October 23. http://abcnews.go.com/US/story?id=95285&page=1.

Selby, Jan. 2014. “Positivist Climate Conflict Research: A Critique.” Geopolitics 19 (4): 829–856.

Semetko, Holli A. and Patti M. Valkenburg. 2000. “Framing European Politics: A Content Analysis of Press and Television News.” Journal of Communication 50 (2): 93–109.

Sena, Lelisa and Kifle W. Michael. 2006. Disaster Prevention and Preparedness. Jimma, Ethiopia: EPHTI.

Shanahan, Elizabeth A., Mark K. McBeth, and Paul L. Hathaway. 2011. “Narrative Policy Framework: The Influence of Media Policy Narratives on Public Opinion.” Politics & Policy 39 (3): 373–400.

Sharp, Gene. 2005. Waging Nonviolent Struggle: 20th Century Practice and 21st Century Potential. Boston, MA: Porter Sargent.

Shebala, Marley. 2009. “Poison in the Earth. 1979 Church Rock Spill a Symbol for Uranium Dangers.” Navajo Times, July 23. http://navajotimes.com/news/2009/0709/072309 uranium.php.

Shoemaker, Pamela J. 1984. “Media Treatment of Deviant Political Groups.” Journalism Quarterly 61: 66–75.

Simmons, Erica. “Grievances Do Matter in Mobilization.” Theory and Society (2014) 43: 513– 546.

167

Sindermann, Carl. 2006. Coastal Pollution. Effects on Living resources and Humans. Boca Raton, FL: Taylor and Francis.

Slettebak, Rune. 2012. “Don’t Blame the Weather! Climate-related Natural Disasters and Civil Conflict.” Journal of Peace Research 49 (1): 163–176.

Slovic, Paul. 1987. “Perception of Risk.” Sciences 236: 80–85.

______. 2000. The Perception of Risk. Routledge.

Slovic, Paul and Elke U. Weber. 2002. “Perception of Risk Posed by Extreme Events.” Paper prepared for the conference “Risk Management strategies in an Uncertain World,” Palisades, New York, April 12-13.

Slothuus, Rune and Claes H. de Vreese. 2010. “Political Parties, Motivated Reasoning, and Issue Framing Effects.” The Journal of Politics 72 (3): 630–645.

Smith, Charlie. 2014. “Environmentalist Credits Regional District for Filming Tailings Flowing from Mount Polley Mine.” The Georgia Straight, 7 August. http://www.straight.com/news/703411/environmentalist-credits-regional-district-filming- tailings-flowing-mount-polley-mine.

Smith, Conrad. 1993. “News Sources and Power Elites in News Coverage of the Exxon Valdez Oil Spill.” Journalism Quarterly 70 (2): 393–403.

Sniderman, Paul M., and Sean M. Theriault. 2004. “The Structure of Political Argument and the Logic of Issue Framing.” In Studies in Public Opinion, edited by Willem E. Saris, and Paul M. Sniderman, 133–165. Princeton, NJ: Princeton University Press.

Soderqvist, Tore. 2000. “Natural Resources Damage from Chernobyl: Further Results.” Environmental and Resource Economics 16: 343–346.

Somers, R. 2004. “Southern Company Builds an Enterprise GIS Framework.” Geospatial Solutions 14 (4).

Soroka, Stuart Neil. 2002. Agenda-setting Dynamics in Canada. Vancouver: UBC Press.

Sovacool, Benjamin K. 2008. “The Costs of Failure: A Preliminary Assessment of Major Energy Accidents, 1907–2007.” Energy Policy 36 (5): 1802–1820.

______. 2010. “A Critical Evaluation of Nuclear Power and Renewable Electricity in Asia.” Journal of Contemporary Asia 40 (3): 369–400.

168

______. 2011. Contesting the Future of Nuclear Power: A Critical Global Assessment of Atomic Energy. Singapore: World Scientific.

Starr, Harvey. 1991. “Democratic Dominoes: Diffusion Approaches to the Spread of Democracy.” Journal of Conflict Resolution 35: 6–31.

Stempel, Guido H. and B. Westley, eds. 1989. Research Methods in Mass Communication, 2nd Ed. Englewood Cliffs, NJ: Prentice Hall.

Stonich, Susan C. 1998. “The Political Ecology of Tourism.” Annals of Tourism Research 25: 25–54.

Sui, Danile Z. and Peter Hugill. 2002. “A GIS-based Spatial Analysis on Neighborhood Effects and Voter Turnout: A Case Study in College Station, Texas.” Political Geography 21(2): 159–173.

Suliman, Samid. 2014. “Protest.” Global Discourse 4 (2-3): 109–119.

Swedish Environmental Protection Agency. Sweden’s Environment Problems and Protection 1960–2010. Stockholm: SEPA.

Tanaka, Shunichi. 2012. “Accident at the Fukushima Daiichi Nuclear Power Stations of TEPCO —Outline & lessons learned.” Proceedings of the Japan Academy, Series B Physical and Biological Sciences 88 (9): 471–484.

Tannen D, ed. 1993. Framing in Discourse. New York: Oxford University Press.

Tarrow, Sidney. 1994. Power in Movement: Social Movements and Contentious Politics. New York: Cambridge University Press.

______. 1996. “States and Opportunities: The Political Structuring of Social Movements.” In Comparative Perspectives on Social Movements, Political Opportunities, and Cultural Framings, edited by Doug McAdam, John McCarthy and Mayer Zald. Cambridge: Cambridge University Press.

______. 2001. “Transnational Politics: Contention and Institutions in International Politics.” Annual Review of Political Science 4: 1–20.

______. 2011. Power in Movement: Social Movements and Contentious Politics. 3rd ed. Cambridge: Cambridge University Press.

169

Taylor, Alan. 2011. “A Flood of Red Sludge, One Year Later.” The Atlantic, 28 September. https://www.theatlantic.com/photo/2011/09/a-flood-of-red-sludge-one-year- later/100158/.

Taylor, Shelley. 1991. “Asymmetrical Effects of Positive and Negative Events: The Mobilization-Minimization Hypothesis.” Psychological Bulletin 110: 67–85.

Tennessee Valley Authority. 2015. Kingston Fossil Plant Coal Ash Release Natural Resource Damage Assessment. The State of Tennessee.

Tilly, Charles. 1978. From Mobilization to Revolution. New York: Random House.

______. 2004. Social Movements 1768-2004. Boulder: Paradigm Publishers.

The Associated Press. 2010. “Brown Pelican, Louisiana's State Bird, Imperiled by Gulf of Mexico Oil Spill.” Times-Picayune, 30 April. https://www.newscientist.com/blogs/shortsharpscience/2010/04/the-worst-environmental- disast.html.

The Economist. 2014. “Japan’s Anti-nuclear Movement. Where’s the Protest?” The Economist, 3 August. http://www.economist.com/blogs/banyan/2014/08/japan-s-anti-nuclear- movement.

The Guardian. 2011. “Nuclear Power Plant Accidents: Listed and Ranked since 1952.” The Guardian, 14 March. https://www.theguardian.com/news/datablog/2011/mar/14/nuclear- power-plant-accidents-list-rank#data.

The New York Times. 1977. “Tanker Explodes In 50‐Mile Slick West of Honolulu.” February 25. http://www.nytimes.com/1977/02/25/archives/tanker-explodes-in-50mile-slick-west- of-honolulu.html.

Thomashaw, M. 1995. Ecological Identity: Becoming a Reflective Environmentalist. Cambridge, MA: MIT Press.

Trumbo, Craig. 1996. “Constructing Climate Change: Claims and Frames in US News Coverage of an Environmental Issue.” Public Understanding of Science 5: 269–283.

Turner, Matthew D. 2003. “Methodological Reflections on the Use of Remote Sensing and Geographic Information Science in Human Ecological Research.” Human Ecology 31 (2): 255–279.

Tvedt, Terje. 1998. Angels of Mercy or Development Diplomats? NGOs and Foreign Aid. Oxford, UK: James Currey.

170

Tversky, Amos and Daniel Kahneman. 1981. “The Framing of Decisions and the Psychology of Choice.” Science 211 (4481): 453–458.

______. 1986 “Rational Choice and the Framing of Decisions.” The Journal of Business 59 (4): S251–S278.

______. 1991. “Loss Aversion in Riskless Choice: A Reference-Dependent Model.” The Quarterly Journal of Economics 106 (4): 1039–1061.

______. 1992. “Advances in Prospect Theory: Cumulative Representation of Uncertainty.” Journal of Risk and Uncertainty 5: 297–323.

Tversky, Amos and Eldar Shafir. 1992. “The Disjunction Effect in Choice under Uncertainty.” Psychological Science 3 (5): 305–309.

UNEP/OCHA. 2000. Spill of Liquid and Suspended Waste at the Aurul S.A. Retreatment Plant in Baia Mare. Geneva: UNEP.

UNISDR. 2004. Living with Risk. A Global Review of Disaster Reduction Initiatives. New York and Geneva: UNISDR.

Urkidi, Leire and Mariana Walter. 2011. “Dimensions of Environmental Justice in Anti-Gold Mining Movements in Latin America.” Geoforum 42: 683–695.

US Coast Guard. 1993. T/V Exxon Valdez Oil Spill. Federal on Scene Coordinator’s Report. Vol. 1. Washington, DC: Department of Transportation.

USDA West Virginia. “The Buffalo Creek Flood Logan County West Virginia February 26 1972.” Association of State Dam Safety Officials. http://www.damsafety.org/media/Documents/More%20Images/Dam%20Failures/Buffalo Creek_SCSpresentation.pdf.

Van Stekelenburg, Jacquelien and Bert Klandermans. 2013. “The Social Psychology of Protest.” Current Sociology Review 61 (5-6): 886–905.

Vesalon, Lucian and Remus Creţan. 2013. “Cyanide Kills!’ Environmental Movements and the Construction of Environmental Risk at Roşia Montană, Romania.” Area 45 (4): 443–451.

Vinnell, Lauren Jennifer, John McLure, and Taciano Lemos Milfont. 2017. “Do Framing Messages Increase Support for Earthquake Legislation?” Disaster Prevention and Management 26 (1): 28–40.

171

Visser, Nick. 2017. “Thousands Fought Against The Dakota Access Pipeline. Now It's Set To Flow Oil.” Huffington Post. 6 January. https://www.huffingtonpost.ca/entry/dakota- access-pipeline-protest-photos_us_592faa01e4b0540ffc847a58.

Vucetich, John A., Jeremy T. Bruskotter, and Michael Paul Nelson. 2015. “Evaluating Whether Nature’s Intrinsic Value Is an Axiom of Or Anathema to Conservation.” Conservation Biology 29 (2): 321–332.

Wallace, Jeremy. 2013. “Cities, Redistribution, and Authoritarian Survival.” Journal of Politics 75 (3): 632–645.

Walsh, Edward J. 1981. “Resource Mobilization and Citizen Protest in Communities around Three Mile Island.” Social Problems 29 (1): 1–21.

Wapner, Paul. 1996. Environmental Activism and World Civic Politics. Albany: State University of New York.

Watts, Michael J. 2005. “Righteous Oil? Human Rights, the Oil Complex and Corporate Social Responsibility.” Annual Review of Environment and Resources 30: 373–407.

Watson, D., L. A. Clark, and A. Tellegen. 1988. “Development and Validation of Brief Measures of Positive and Negative Affect: The PANAS Scales.” Journal of Personality and Social Psychology 54: 1063–1070.

Waugh, William, Jr. 2006. “The Political Costs of Failure in the Katrina and Rita Disasters.” The Annals of the American Academy of Political and Social Science 604: 10–25.

Weaver, R. Kent. 1986. “The Politics of Blame Avoidance.” Journal of Public Policy 6: 371– 398.

Weeks, Brian. 2015. “Emotions, Partisanship, and Misperceptions: How Anger and Anxiety Moderate the Effect of Partisan Bias on Susceptibility to Political Misinformation.” Journal of Communication 65: 699–719.

Weersink, Alfons and Mark Raymond. 2007. Environmental Regulations Impact on Agricultural Spills and Citizen Complaints.” Ecological Economics 60: 654–660.

Weisberg, Вarbara. 1993. “One City's Approach to NIMBY: How New York City Developed a Fair Share Siting Process.” Journal of the American Planning Association 59: 93–97.

Welzel, Christian and Franziska Deutsch. 2011. “Emancipative Values and Non-Violent Protest: The Importance of ‘Ecological’ Effects.” British Journal of Political Science 42: 465– 479.

172

West, David. 2013. Social Movements in Global Politics. Cambridge, UK: Polity Press.

Wexler, Mark N. 1996. “A Sociological Framing of the NIMBY (Not-In-My-Backyard) Syndrome.” International Review of Modern Sociology 26 (1): 91–110.

Wheaton, Sarah. 2010. “Protesters Gather at BP Gas Stations.” The New York Times, 2 June. http://www.nytimes.com/2010/06/03/us/03boycott.html.

White, Helen, Pen-Yuan Hsing, Walter Cho, Timothy M. Shank, Erik E. Cordes, Andrea M. Quattrini, Robert K. Nelson, Richard Camilli, Amanda W. J. Demopoulos, Christopher R. German, James M. Brooks, Harry H. Roberts, William Shedd, Christopher M. Reddy, and Charles R. Fisher. 2012. “Impact of the Deepwater Horizon Oil Spill on a Deep- water Coral Community in the Gulf of Mexico.” Proceedings of the National Academy of Sciences of the United States of America 109 (50): 20303–20308.

Wiklund, Jan. “The Swedish Environmental Movements.” Folkrorelser och Protester. http://www.folkrorelser.org/inenglish/swedishenviron.html.

Wilder, Andrew. 2010. “Aid and Stability in Pakistan: Lessons from the 2005 Earthquake Response.” Disasters 34 (3): S406–S426.

Windsor, Leah C., Nia Dowell, and Art Grasser. 2014. “The Language of Autocrats: Leaders’ Language in Natural Disaster Crises.” Risk, Hazards, & Crisis in Public Policy 5 (4): 446–467.

WISE. 2017. “Chronology of Major Tailings Dam Failures.” WISE Uranium Project, 3 April. http://www.wise-uranium.org/mdaf.html.

Wolsink, Maarten. 1994. “Entanglement of Interests and Motives: Assumptions behind the NIMBY Theory on Facility Siting.” Urban Studies 31: 851–866.

Wood, Elisabeth. 2003. Insurgent Collective Action and Civil War in El Salvador. Cambridge: Cambridge University Press.

Wood, Rheed and Torin Wright. 2015. “Responding to Catastrophe: Repression Dynamics Following Rapid-onset Natural Disasters.” Journal of Conflict Resolution: 1–27.

World Nuclear Association. 2017. “Fukushima Accident.” April. http://www.world- nuclear.org/information-library/safety-and-security/safety-of-plants/fukushima- accident.aspx.

173

______. 2012. “Three Mile Island Accident.” January. http://www.world- nuclear.org/information-library/safety-and-security/safety-of-plants/three-mile-island- accident.aspx.

______. 2013. “Tokaimura Criticality Accident 1999.” October. http://www.world- nuclear.org/information-library/safety-and-security/safety-of-plants/tokaimura-criticality- accident.aspx.

______. 2016. “Fukushima: Radiation Exposure”. February. http://www.world- nuclear.org/information-library/safety-and-security/safety-of- plants/appendices/fukushima-radiation-exposure.aspx.

Wridt, Pamela. 2010. “A Qualitative GIS Approach to Mapping Urban Neighborhoods with Children to Promote Physical Activity and Child‐friendly Community Planning.” Environment and Planning B: Planning and Design 37: 129–147.

Xiaoa, Jieying, Yanjun Shen, Jingfeng Ge, Ryutaro Tateishia, Changyuan Tang, Yanqing Liang, and Zhiyying Huang. 2006. “Evaluating Urban Expansion and Land Use Change in Shijiazhuang, China, by Using GIS and Remote Sensing.” Landscape and Urban Planning 75, No. 1-2: 69–80.

Yle News. 2012. “’Stop Talvivaara’ protest brings carnival mood to Helsinki.” 14 November. https://yle.fi/uutiset/osasto/news/stop_talvivaara_protest_brings_carnival_mood_to_helsi nki/6376956.

Zald, Mayer N. 1996. “Culture, Ideology, and Strategic Framing.” In Comparative Perspectives on Social Movements, edited by Doug McAdam, John D. McCarthy, and Mayer N. Zald, 261–274. New York: Cambridge University Press.

Zald, Mayer and John McCarthy. 1979. The Dynamics of Social Movements. Cambridge: Winthrop.

174

Appendix 1

Table 30. Disaster locations and areas impacted.

ID Event Pollution Location Area Impacted Source Type Coordinates

1 Kalamazoo heavy 42.25743 35 mi (56 km) down the EMC 2010 River bitumen river toward Lake Michigan -84.99307

2 Deepwater crude 28.736667 More than 1,100 mi (1,770 Aigner et al. 2010; Horizon km) of coastline, at least NRDC 2015 -88.387167 1,200 sq. mi (3,100 km²) of the deep ocean floor, and 68,000 sq. mi (176,000 km²) of surface water

3 Prestige heavy fuel 42.316667 Total of 30,000 km²; spread Balseiro et al. 2003; oil toward the Cantabric Sea Montero et al. 2003; -12.316667 and the Bay of Biscay along Munilla et al. 2011 2,600 coastal km, reaching the Bay of Arcachon (France)

4 Sea Empress light crude 51.6796 Over 200 km of shoreline Law and Kelly 2004

-5.1777

5 MV Braer light crude 59.883333 An 8-km radius Forbes and Campbell 1993; Fisheries -1.35 Research Services 2003

6 Aegean Sea light crude 43.389 Northeast across 200 km of Pastor et al. 2001 shoreline (including -8.41 estuaries of Ares and el Ferrol)

7 MT Haven heavy 44.22 The Ligurian coast, from Madrid et al. 2015 crude Genoa to Savona 8.46

8 Exxon Valdez crude 60.838833 Total of 1,300 mi (2,000 US Coast Guard 1993; km); from Bligh Reef spread Holba 2014 -146.883 460 mi (740 km) to Chignik on the Alaska Peninsula

9 Odyssey North Sea 39.47012 A 3 mi (5 km) by 10 mi (16 CEDRE100 crude km) area -48.86718

100 The CEDRE database of oil spills is available at http://wwz.cedre.fr/en/Our-resources/Spills/(letter)/default.

ID Event Pollution Location Area Impacted Source Type Coordinates

10 Irenes light crude 36.916667 The oil slick was 2 mi (3 Hellenic Shipping Serenade km) long and 0.5 mi (0.8 News 2016; ITOPF101 21.7 km) wide, with pockets of oil in the Navarino Bay (along shoreline); up to 100 km of shoreline oiled

11 Independenta Libyan 41.033333 The port of Haydrapasa CEDRE; NOAA crude Incident News102 28.95

12 Atlantic crude 11.5 The AC: oil slick was about Horn and Neal 1981 Empress/ 10 mi (16 km) long and 2 mi -60.533333 (3 km) wide. The AE: oil Aegean slick covered about 2 mi (3 Captain km) by 15 mi (24 km). The oil was 10 mi (16 km) from the north coast of Tobago.

13 Betelgeuse mixed 50.666667 To east shore of Bantry Bay NOAA Incident News, Arabian and Reenydonagan Point on “Behaviour of Oil” crude -12.066667 Whiddy Island, then to the north and south shores of Bantry Bay, reaching Castle Townbere on the north shore of the bay, and League Point on the south shore. Bear Island was also impacted.

14 Andros Patria Iranian 45.5 Two oil slicks and about 100 CEDRE heavy m of oil on the shore crude -4.333333

15 Amoco Cadiz crude 48.583333 29 km wide and 129 km NOAA Incident News long; covered 320 km of -4.716667 shoreline from Brest to Saint Brieuc

16 Hawaiian Indonesian 21.3917 Oil slick 50 miles (80 km) The New York Times Patriot crude long 1977; Duffy and Elliott -162.57568 2010

101 ITOPF database is available at http://www.itopf.com/in-action/recent-case-studies/. 102 The NOAA Incident News database is available at https://incidentnews.noaa.gov/.

176

ID Event Pollution Location Area Impacted Source Type Coordinates

17 Urquiola light crude 43.36667 Total of 215 km of the NOAA Incident News, northwestern Spanish; Rias “Behaviour of Oil” 8.383333 de La Coruna, Rias Area, and Rias Betanzos, beaches at Raso and Perbes, reaching Playa de Doninos, approximately 16 km from the grounding site. Further north near Frouxeria and Pantin and offshore of Ria de Cedeira. Southwest near Barranan. Westward contamination stretched to Playa de Beo, 45 km from the grounding.

18 Jakob Maersk Iranian 41.15 15 to 20 km north and south NOAA Incident News crude of the wreck site, 19 mi (30 -8.833333 km) of coastline

19 Othello heavy fuel 59.43333 30 km of coastline CEDRE oil -18.38333

20 Torrey Canyon Kuwait 50.05 Three distinct slicks: the first NOAA Incident News crude drifted up the English -4.733333 Channel, oiling the north

coasts of France and Guernsey. The second covered 200 mi (320 km) of the coast of West Cornwall, and 100 mi (160 km) of coastline between Perranporth and The Lizard, at the southern tip of Cornwall. The third drifted south into the Bay of Biscay.

21 Lakeview crude 35.09142 Between the towns of Taft Li and Molin 2014, p. Gusher and Maricopa; 60 acres 179 -119.4014 (0.24 km²) “lake of oil” near the site

177

ID Event Pollution Location Area Impacted Source Type Coordinates

22 Mount Polley tailings 52.51333 Polley Lake to Hazeltine Birchwater 2014 slurry, Creek (grew from 2 m to 50 process -121.5964 m across) and Quesnel Lake water

23 Talvivaara aluminium, 63.97166 At least 100 hectares (1 km²) Rweyendela 2014 cadmium, of marshland, streams, lakes nickel, 28.005 and ponds (including the uranium, northern Oulujoki-waterway zinc and to the south through the lake Yla-Lumijarvi in the major East Finland Vuoksi waterway)

24 Kingston coal fly 35.8982 The Emory River and its Tennessee Valley Fossil Plant ash slurry Swan Pond embayment, on Authority 2015; EPA -84.5188 to the opposite shore, covering up to 300 acres (1.2 km²) of the surrounding land; 9 km up Emory River, and 900 km down the Clinch River.

25 Martin County coal slurry 37.87959 Two tributaries of the Tug Kilborn 2000; Sealey Fork River; Wolf Creek on 2000; Scott et al. 2005; -82.60861 Coldwater Fork (grew from Lovan 2010 3 m to 91 m across); 300 – 500 km of the Big Sandy River and its tributaries and the Ohio River

26 Aitik mine waste 67.066667 the Leipojoki River, the European Commission and earth 20.95 Sakajoki River fill

27 Baia Mare cyanide 47.70639 The Lapus and Somes UNEP/OCHA 2000; tributaries of the river Tisza, European Environment 23.68417 then Danube Agency 2009

28 Los Frailes tailings 37.516667 River Agrio, then River Pain, Sanchez, and (Doñana solids and Guadiamar for about 45 km; Meharg 1998 disaster) water -6.25 affected some 4,400 ha of crops, pasture and woodland, and 800 ha of river and marshland as well as 1,290 ha of National and Natural Parks

29 Tyrone tailings 32.641806 8 km downstream Earthworks 2012 slurry -108.325417

178

ID Event Pollution Location Area Impacted Source Type Coordinates

30 Fukushima 37.4213 Total of 1000 km²; 20 km World Nuclear radius evacuation zone, main Association 2016 and 141.0281 concentration stretched 2017 northwest from the plant out to 40 km away

31 Tokaimura 36.4664 10 km radius, 350 m radius World Nuclear evacuation zone Association 2013 140.6067

32 Saint Laurent 47.72 No data on spread des Eaux 1.5775

33 Church Rock 35.650833 The Puerco River – 80 miles Shebala 2009 (130 km) downstream; -108.506389 groundwater contamination

34 Three Mile 40.1547 10 mi (16 km) radius, 5 mi Battist and Peterson; Island (8 km) radius evacuation Nuclear Safety -76.7252 zone Analysis Center 1980; World Nuclear Association 2012

35 Lucens reactor 46.6928 The underground cavern Pinto 1979 containing the reactor, 17 m 6.827 in diameter

36 SL-1 43.5195 500 feet (152 m) would have Joint Committee on been the evacuation zone but Atomic Energy 1961 -112.0458 the reactor was not in a populated area; a large plume moved southeast to the edges of the towns of Burley and Rupert (about 100 km away)

37 Windscale fire 54.42158 The most heavily Loutit, Marley, and contaminated area was 200 Russell 1960; Nelson, -3.499997 square miles (518 km²), but Kitchen, and Maryon the radiation extended into 2006 Europe (as far as Frankfurt).

38 Chalk River 46.050242 No data on spread

-77.361002

179

Table 31. Post-disaster protest events.

Event Protest type Date Location Size Size Source accuracy

Deepwater demonstration May 24, Houston 50 Sourced Christian Science Horizon 2010 Monitor, June 12, 2010; Houston Chronicle, May 24, 2010

Deepwater petition June 1, USA 21000 Sourced Christian Science Horizon 2010 Monitor, June 12, 2011

Deepwater demonstration May 1, Clearwater, FL 50 Sourced St. Petersburg Horizon 2010 Times, June 1, 2010

Deepwater sabbotage/ July 27, London, UK 10 Estimate Business World, July 27, Horizon 2010 (Small) 2010 activist stunt

Deepwater demonstration May 1, New York, NY 200 Sourced The Daily Telegraph, Horizon 2010 June 5, 2010

Deepwater sabbotage/ July 13, London, UK 10 Estimate The Daily Telegraph, Horizon 2010 (Small) July 14, 2010 activist stunt

Deepwater activist stunt June 1, London, UK 10 Estimate The Guardian, June 24, Horizon 2010 (Small) 2010

Deepwater activist stunt July 1, Nottingham, UK 10 Estimate Nottingham Evening Horizon 2010 (Small) Post, July 13, 2010

Deepwater activist stunt May 1, London, UK 10 Estimate The Gold Coast Bulletin, Horizon 2010 (Small) May 21, 2010

Deepwater demonstration May 30, Pantops, VA 15 Sourced The Clover Herald, Horizon 2010 June 1, 2010

Deepwater demonstration May 1, Daytona Beach 100 Sourced News-Journal (Daytona Horizon 2010 Shores, FL Beach, Florida), June 17, 2010

Deepwater online Aug 1, USA 849000 Sourced The Washington Post, Horizon petition/campa 2010 August 21, 2010 ign

Deepwater activist stunt July 30, UC Berkeley 10 Estimate Morning Star, August 2, Horizon 2010 (Small) 2010

Prestige demonstration Dec 15, Barcelona, Spain 50000 Sourced The Guardian, 2002 (Range) December 16, 2002

Prestige demonstration Dec 1, Santiago de 200000 Sourced The Times, December 2, 2002 Compostela, 2002 Spain

180

Event Protest type Date Location Size Size Source accuracy

Prestige demonstration Feb 1, Madrid, Spain 100000 Sourced Daily Post (North 2003 Wales), February 24, 2003

Sea petition Apr 1, UK 100000 Sourced The Independent Empress 1996 (London), April 25, 1996

Sea sabbotage/ Apr 1, Milford Haven, 10 Estimate The Independent Empress 1996 Wales (Small) (London), April 25, 1996 activist stunt

Aegean demonstration Jan 10, La Coruna, 10000 Sourced The Herald (Glasgow), Sea 1993 Spain January 11, 1993

Exxon sabbotage Apr 1, St. Loius, MO 1 Sourced St. Louis Post-Dispatch, Valdez 1989 April 16, 1989

Exxon demonstration Apr 1, Alki Beach, 2000 Sourced St. Louis Post-Dispatch, Valdez 1989 Seattle, WA April 16, 1989

Exxon demonstration July 1, Tidelands Park, 100 Sourced St. Louis Post-Dispatch, Valdez 1989 Coronado, CA April 16, 1989

Exxon boycott May 1, USA 10000 Sourced The New York Times, Valdez 1989 July 12, 1989

Exxon activist stunt Apr 1, Anchorage, 1 Sourced St. Louis Post-Dispatch, Valdez 1989 Alaska May 3, 1989

Exxon demonstration May 1, Parsippany, NJ 250 Sourced St. Louis Post-Dispatch, Valdez 1989 April 30, 1989

Exxon activist stunt Sept 1, Valdez harbor, 50 Sourced The New York Times, Valdez 1989 Alaska May 19, 1989

Exxon demonstration May 1, Anchorage, 400 Sourced The New York Times, Valdez 1989 Alaska September 10, 1989

Exxon boycott Apr 1, USA 1000 Sourced St. Louis Post-Dispatch, Valdez 1989 May 3, 1989

Amoco demonstration Mar 1, France 15000 Sourced The New York Times, Cadiz 1978 April 17, 1989

Amoco demonstration Mar 27, Portsall, France 2000 Sourced The New York Times, Cadiz 1978 March 31, 1978

Fukushima demonstration Mar 12, Germany: 45km 60000 Sourced The Globe and Mail, 2011 between March 29, 1978 Stuttgart and the Neckarwestheim nuclear power plant

Fukushima demonstration Mar 26, Berlin, Germany 110000 Sourced Hasegawa (2014) 2011

181

Event Protest type Date Location Size Size Source accuracy

Fukushima demonstration Mar 26, Hamburg, 45000 Sourced San Jose Mercury News, 2011 Germany March 26, 2011

Fukushima demonstration Mar 26, Munich, 35000 Sourced San Jose Mercury News, 2011 Germany March 26, 2011

Fukushima demonstration Mar 26, Cologne, 40000 Sourced San Jose Mercury News, 2011 Germany March 26, 2011

Fukushima demonstration Apr 10, Tokyo, Japan 15000 Sourced San Jose Mercury News, 2011 March 26, 2011

Fukushima demonstration June 11, Tokyo, Japan 20000 Sourced Hasegawa (2014) 2011

Fukushima demonstration Sept 19, Tokyo, Japan 64000 Sourced Hasegawa (2014) 2011

Fukushima demonstration Apr 30, Taipei, Taiwan 10000 Sourced Hasegawa (2014); Elliott 2011 (2013)

Fukushima demonstration Apr30, Kaohsiung, 5000 Sourced BBC Monitoring Asia 2011 Taiwan Pacific, April 30, 2011

Fukushima demonstration Apr 30, Taitung, Taiwan 1000 Sourced BBC Monitoring Asia 2011 Pacific, April 30, 2011

Tokaimura petition Oct 1, Japan 2170 Sourced BBC Monitoring Asia 1999 Pacific, April 30, 2011

Three Mile demonstration May 6, Washington, DC 107500 Sourced The New York Times, Island 1979 January 13, 2000

Three Mile demonstration Apr 28, Rocky Flats 7700 Sourced The Washington Post, Island 1979 nuclear weapons May 7, 1979 plant, CO

Three Mile demonstration Apr 8, Washington, DC 500 Sourced The Washington Post, Island 1979 April 29, 1979

Mount demonstration Aug 11, Vancouver, BC 10 Estimate The Washington Post, Polley 2014 (Small) April 29, 1979

Talvivaara demonstration Nov 14, Helsinki, Finland 1000 Sourced US Official News, 2012 August 16, 2014

Talvivaara petition Nov 18, Finland 20000 Sourced Yle, November 14, 2012 2012

Talvivaara demonstration Nov 18, Oulu, Finland 10 Estimate Nuclear Heritage 2016 2012 (Small)

Talvivaara demonstration Apr 1, Sotkamo, 100 Sourced Nuclear Heritage 2016 2012 Finland

182

Appendix 2

Table 32. Analysis of necessary conditions for the outcome ‘small protest’.

Condition Consistency Coverage ~ Condition Consistency Coverage

GROWTH 0.000 0.000 not GROWTH 1.000 0.1000 INFL 0.000 0.000 not INFL 1.000 0.1250 INEQ 0.000 0.000 not INEQ 1.000 0.0909 POV 0.000 NA not POV 1.000 0.0588 ENV 1.000 0.0625 not ENV 0.000 0.0000 URBAN 1.000 0.0625 not URBAN 0.000 0.0000 MOVM 1.000 0.0588 not MOVM 0.000 NA IALLY 1.000 0.0625 not IALLY 0.000 0.0000 OCAMP 0.000 0.000 not OCAMP 1.000 0.0833 REPR 0.000 0.000 not REPR 1.000 0.0714 POSTCW 1.000 0.0588 not POSTCW 0.000 NA INDUSTR 1.000 0.0625 not INDUSTR 0.000 0.0000 COMM 0.750 0.1111 not COMM 0.000 0.0000 TRADE 1.000 0.0588 not TRADE 0.000 NA

Table 33. Analysis of necessary conditions for the outcome ‘medium protest’.

Condition Consistency Coverage ~ Condition Consistency Coverage

GROWTH 0.000 0.0000 not GROWTH 1.000 0.400 INFL 0.500 0.2222 not INFL 0.500 0.250 INEQ 0.250 0.1667 not INEQ 0.750 0.273 POV 0.000 NA not POV 1.000 0.235 ENV 1.750 0.1875 not ENV 0.250 1.000 URBAN 1.000 0.2500 not URBAN 0.000 0.000 MOVM 1.000 0.2353 not MOVM 0.000 NA IALLY 1.000 0.2500 not IALLY 0.000 0.000 OCAMP 0.250 0.2000 not OCAMP 0.750 0.250

183

Condition Consistency Coverage ~ Condition Consistency Coverage

REPR 0.000 0.0000 not REPR 1.000 0.286 POSTCW 1.000 0.2353 not POSTCW 0.000 NA INDUSTR 1.000 0.2500 not INDUSTR 0.000 0.000 COMM 0.750 0.3333 not COMM 0.250 0.125 TRADE 1.000 0.2353 not TRADE 0.000 NA

Table 34. Analysis of necessary conditions for the outcome ‘large protest’.

Condition Consistency Coverage ~ Condition Consistency Coverage

GROWTH 0.6667 0.2857 not GROWTH 0.3333 0.1000 INFL 0.6667 0.2222 not INFL 0.3333 0.1250 INEQ 0.3333 0.1667 not INEQ 0.6667 0.1818 POV 0.0000 NA not POV 1.0000 0.1765 ENV 1.0000 0.1875 not ENV 0.0000 0.0000 URBAN 1.0000 0.1875 not URBAN 0.0000 0.0000 MOVM 1.0000 0.1765 not MOVM 0.0000 NA IALLY 1.0000 0.1875 not IALLY 0.0000 0.0000 OCAMP 0.6667 0.4000 not OCAMP 0.3333 0.3333 REPR 0.0000 0.0000 not REPR 1.0000 0.2143 POSTCW 1.0000 0.1765 not POSTCW 0.0000 NA INDUSTR 1.0000 0.1875 not INDUSTR 0.0000 0.0000 COMM 0.6667 0.2222 not COMM 0.3333 0.1250 TRADE 1.0000 0.1765 not TRADE 0.0000 NA

184

Table 35. Truth table for the outcome ‘small protest’.

GROW INFL INEQ ENV URB MOV IALLY OCAMP REPR POSTCW INDUS COM

0 0 0 1 1 1 1 0 0 1 1 0 0 0 0 1 1 1 1 0 0 1 1 1 0 0 1 1 1 1 1 0 0 1 1 1 0 1 0 0 1 1 1 1 0 1 1 0 0 1 0 1 1 1 0 0 0 1 1 0 0 1 0 1 1 1 1 0 0 1 1 1 0 1 0 1 1 1 1 1 0 1 1 1 1 0 0 1 1 1 1 0 1 1 1 1 1 0 0 1 1 1 1 0 0 1 1 1 1 0 0 1 1 1 1 1 1 1 1 0 1 1 0 1 0 1 1 0 1 1 0 0 1 1 1 1 1 1 1 0 1 1 0 0 1 1 1 1 1 1 1 1 0 1 1 0

TRADE OUT N Cons. 1 0 1 0.000 1 0 2 0.500 1 0 2 0.000 1 0 1 0.000 1 0 1 0.000 1 0 1 0.000 1 0 1 0.000 1 0 1 0.000 1 0 2 0.000 1 0 1 0.000 1 0 1 0.000 1 0 1 0.000 1 0 2 0.000

185

Table 36. Truth table for the outcome ‘medium protest’.

GROW INFL INEQ ENV URB MOV IALLY OCAMP REPR POSTCW INDUS COM

0 0 0 1 1 1 1 0 0 1 1 0 0 0 0 1 1 1 1 0 0 1 1 1 0 0 1 1 1 1 1 0 0 1 1 1 0 1 0 0 1 1 1 1 0 1 1 0 0 1 0 1 1 1 0 0 0 1 1 0 0 1 0 1 1 1 1 0 0 1 1 1 0 1 0 1 1 1 1 1 0 1 1 1 0 1 1 1 1 1 1 0 1 1 1 1 1 0 0 1 1 1 1 0 0 1 1 1 1 0 0 1 1 1 1 1 1 1 1 0 1 1 0 1 0 1 1 0 1 1 0 0 1 1 1 1 1 1 1 0 0 1 1 0 1 1 1 1 1 1 1 1 0 1 1 0

TRADE OUT N Cons.

1 0 1 0.000 1 0 2 0.500 1 0 2 0.500 1 1 1 1.000 1 0 1 0.000 1 1 1 1.000 1 0 1 0.000 1 0 1 0.000 1 0 2 0.000 1 0 1 0.000 1 0 1 0.000 1 0 1 0.000 1 0 2 0.000

186

Table 37. Truth table for the outcome ‘large protest’.

GROW INFL INEQ ENV URB MOV IALLY OCAMP REPR POSTCW INDUS COM

0 0 0 1 1 1 1 0 0 1 1 0 0 0 0 1 1 1 1 0 0 1 1 1 0 0 1 1 1 1 1 0 0 1 1 1 0 1 0 0 1 1 1 1 0 1 1 0 0 1 0 1 1 1 0 0 0 1 1 0 0 1 0 1 1 1 1 0 0 1 1 1 0 1 0 1 1 1 1 1 0 1 1 1 0 1 1 1 1 1 1 0 1 1 1 1 1 0 0 1 1 1 1 0 0 1 1 1 1 0 0 1 1 1 1 1 1 1 1 0 1 1 0 1 1 1 1 0 1 1 0 0 1 1 1 1 0 1 1 0 0 1 1 0 1 1 1 1 1 1 1 1 0 1 1 0

TRADE OUT N Cons.

1 0 1 0.000 1 0 2 0.000 1 0 2 0.000 1 0 1 0.000 1 0 1 0.000 1 0 1 0.000 1 1 1 1.000 1 0 1 0.000 1 0 2 0.500 1 0 1 0.000 1 0 1 0.000 1 0 1 0.000 1 0 2 0.500

187

Table 38. Conservative solution for the outcome ‘medium protest’.

Configurations Consistency Raw coverage Unique coverage growth*INFL*env*URBAN*MOVM* 1.000 0. 250 0.250 IALLY*repr*INDUSTR+ growth*INFL*ineq*URBAN*MOVM* 1.000 0.250 0.250 IALLY*ocamp*repr*INDUSTR*COMM ➔ EMERGENCE OF MEDIUM P ROTEST Solution consistency 1.000 Solution coverage 0.500

Notes: *denotes logical AND, while + denotes logical OR. Uppercase letters denote the presence of a condition, while lowercase letters denote the absence of a condition.

Table 39. Conservative solution for the outcome ‘large protest’.

Configurations Consistency Raw coverage Unique coverage growth*INFL*ENV*URBAN*MOVM* 1.000 1.000 0.333 IALLY*OCAMP*repr*INDUSTR*COMM ➔ EMERGENCE OF LARGE PROTEST Solution consistency 1.000 Solution coverage 0.333

Notes: *denotes logical AND. Uppercase letters denote the presence of a condition, while lowercase letters denote the absence of a condition.

188

Appendix 3 Content Analysis – Codebook Notes: If more than one frame/actor is present in a story, multiple values per variable are allowed. When coding actors: • If the story does not refer to a specific actor, code the journalist as the framing actor. • Corporation refers to the company responsible for the disaster. • ‘Activist’ may include religious groups. • Local government includes First Nations representatives if in reference to the chief or some governing body. Code any other First Nations organizations as ‘activist’. • ‘Expert’ includes commissions of experts assembled to investigate the disaster cause. ‘Expert’ could be identified as such by the journalist or self-identified (e.g., in opinion pieces). • ‘Other’ includes industry groups, international organizations (e.g., IAEA), and members of the public (e.g., in opinion pieces) (except when interviews with the public are used as human interest frames by the journalist).

Coding Item Explanation

V1. Story identification number

V2. Source Newspaper name and location

V3. Date Story date: day, month, year

V4. Story uniqueness If duplicate, mark as ‘D’ and include the duplicate story ID: e.g., D(146). If there are more than one duplicates of the same story, only use the ID number of the original/first story.

V5. Primary topic 1 = disaster aftermath 2 = disaster causes 3 = cleanup efforts/disaster response 4 = compensation 5 = protest 6 = other

Note: If the story is fairly balanced in terms of different topics, more than one can be coded.

Disaster aftermath Includes impacts on the people and environment: victims’ suffering (physical, emotional), economic damage (e.g., destruction of property, layoffs), environmental damage (destruction of natural environment).

189

Disaster causes Discussion of what and/or who caused the event. Includes regulatory failures (i.e., ineffective pre-existing regulations, rules, laws, etc.).

Cleanup efforts/disaster Disaster response and/or cleanup efforts by government, corporation response and/or communities. Includes discussions of cost and responsibility for cleanup.

Compensation Lawsuits, fines or any compensation requests (granted or not) linked to the disaster.

Protest Non-violent protest activities explicitly linked to the disaster (e.g., demonstration, petition, boycott, activist stunts). Must be explicitly stated that protest occurred. Simple note of activist or public disagreement does not qualify.

Other Anything else related to the disaster not captured by the other categories (e.g., discussion of new policies or other ways forward, political discussions triggered by the disaster, etc.).

V6. Attribution of Based on answers to questions in Table 1. Code as (1) if the answer is Responsibility ‘yes’. Code as (0) if the answer is ‘no’.

V6a. Framing actor A Subject assigning blame for the disaster. Code as (0) if V6 is absent. (assignment of blame) 1 = Journalist 2 = Activist 3 = Local government 4 = Provincial/state government 5 = Federal/national government 6 = Corporation 7 = Expert 8 = Other 9 = Unclear

Note: Q1d refers to pre-existing practices in terms of regulations or established disaster preparedness/management practices, not to actors’ response in the disaster aftermath.

V6b. Framing actor B Subject held accountable for the disaster. Code as (0) if V6 is absent. (denial of blame) 1 = Journalist 2 = Activist 3 = Local government 4 = Provincial/state government 5 = Federal/national government 6 = Corporation 7 = Expert 8 = Other 9 = Unclear

190

V7. Human interest Based on answers to questions in Table 1. Code as (1) if the answer is ‘yes’. Code as (0) if the answer is ‘no’.

Note: Q2a refers to specific groups of people – code as (1) if article mentions a specific number of people in a specific area, but code as (0) if uses vague terms such as “thousands of people affected”

V7a. Framing actor Subject using the human interest frame. Code as (0) if V7 is absent. If ‘human interest’ stories are presented by the journalist (e.g., if individuals are interviewed), code ‘journalist’ as the framing actor. Only include members of the public in ‘other’ (or other relevant category) if coding opinion pieces or interview transcripts. 1 = Journalist 2 = Activist 3 = Local government 4 = Provincial/state government 5 = Federal/national government 6 = Corporation 7 = Expert 8 = Other 9 = Unclear

V8. Morality Based on answers to questions in Table 1. Code as (1) if the answer is ‘yes’. Code as (0) if the answer is ‘no’.

V8a. Framing actor Subject referring to morals/ethics within the disaster context. Code as (0) if V8 is absent. 1 = Journalist 2 = Activist 3 = Local government 4 = Provincial/state government 5 = Federal/national government 6 = Corporation 7 = Expert 8 = Other 9 = Unclear

V9. Economic Based on answers to questions in Table 1. Code as (1) if the answer is consequences ‘yes’. Code as (0) if the answer is ‘no’.

V9a. Framing actor Subject referring to economic consequences of the disaster. These include harm or negative externalities that are caused by the disaster or subsequent regulations that directly stem from the disaster. These also include harm or negative externalities to local, regional or national economies that rely on the environment. 1 = Journalist 2 = Activist 3 = Local government

191

4 = Provincial/state government 5 = Federal/national government 6 = Corporation 7 = Expert 8 = Other 9 = Unclear

V10. Conflict Based on answers to questions in Table 1. Code as (1) if the answer is ‘yes’. Code as (0) if the answer is ‘no’.

Note: If Q5a is (1), there does not have to be an accusing and an accused actor. The story may simply reflect disagreements among different actors. Existence of public protest, for example, suggests conflict. In Q5b and Q5c, the conflict between specific actors must be explicitly stated.

V10a. Framing actor A Subject making accusations, assigning dramatic labels or arguing in (accusing) dichotomies. Code as (0) if V10 is absent. 1 = Journalist 2 = Activist 3 = Local government 4 = Provincial/state government 5 = Federal/national government 6 = Corporation 7 = Expert 8 = Other 9 = Unclear

V10b. Framing actor B Actor subjected to accusations and/or dramatic labels. Code as (0) if V10 (accused) is absent. 1 = Journalist 2 = Activist 3 = Local government 4 = Provincial/state government 5 = Federal/national government 6 = Corporation 7 = Expert 8 = Other 9 = Unclear

V11. Environmental Based on answers to questions in Table 1. Code as (1) if the answer is damage ‘yes’. Code as (0) if the answer is ‘no’.

V11a. Framing actor Subject referring to the environmental damage (immediate and/or future harm) from the disaster. This includes public health issues (e.g., pollution of drinking water). Environment refers to natural resources, wildlife, air, land, water, and landmarks. 1 = Journalist 2 = Activist

192

3 = Local government 4 = Provincial/state government 5 = Federal/national government 6 = Corporation 7 = Expert 8 = Other 9 = Unclear

V12. Industry-specific Based on answers to questions in Table 1. Code as (1) if the answer is ‘yes’. Code as (0) if the answer is ‘no’.

V12a. Pro-industry Subject defending and/or supporting the industry linked to the disaster. Code as (0) if V12 is absent. 1 = Journalist 2 = Activist 3 = Local government 4 = Provincial/state government 5 = Federal/national government 6 = Corporation 7 = Expert 8 = Other 9 = Unclear

V12b. Anti-industry Subject opposing the industry linked to the disaster. Code as (0) if V12 is absent. 1 = Journalist 2 = Activist 3 = Local government 4 = Provincial/state government 5 = Federal/national government 6 = Corporation 7 = Expert 8 = Other 9 = Unclear

V12c. Indifferent/neutral Subject referring to the industry linked to the disaster in a neutral/indifferent tone. Code as (0) if V12 is absent. 1 = Journalist 2 = Activist 3 = Local government 4 = Provincial/state government 5 = Federal/national government 6 = Corporation 7 = Expert 8 = Other 9 = Unclear

193

V13. Tone: disaster Tone of coverage regarding the government’s or corporation’s handling of management the disaster, including government’s course of action towards the responsible corporation. If the tone is present, the story should convey the sense of the situation being handled well or not well – for example, through emphasizing that proper procedures were followed/established (or not) after the disaster, or through referring to the speed of response (where fast=successful; slow=unsuccessful). Code as ‘neutral’ if the tone is apparent in the story and is fairly balanced (i.e., no positive or negative tone prevailing) – e.g., if the story mentions someone following some procedures, but does not clearly state whether those procedures were good or bad. 0 = Tone absent 1 = Unsuccessful 2 = Successful 3 = Neutral 4 = Unclear

V13a. Framing actor Subject using the tone. Code as (0) if V13 is absent. 1 = Journalist 2 = Activist 3 = Local government 4 = Provincial/state government 5 = Federal/national government 6 = Corporation 7 = Expert 8 = Other 9 = Unclear

V14. Tone: disaster Tone of coverage regarding disaster preparedness – government’s, preparedness corporation’s or community’s. If the tone is present, the story should convey the sense of the relevant actor(s) being prepared for the disaster (or similar disasters) – for example, through emphasizing that pre-existing regulations/procedures were adequate (or not). This tone is specifically about referring to relevant pre-existing regulations or procedures. If, for example, the framing actor refers to weak regulations, the tone conveys lack of preparedness. Code as ‘neutral’ if the tone is apparent in the story and is fairly balanced (i.e., no positive or negative tone prevailing). 0 = Tone absent 1 = Unprepared 2 = Prepared 3 = Neutral 4 = Unclear

V14a. Framing actor Subject using the tone. Code as (0) if V15 is absent. 1 = Journalist 2 = Activist

194

3 = Local government 4 = Provincial/state government 5 = Federal/national government 6 = Corporation 7 = Expert 8 = Other 9 = Unclear

V15. Tone: actor Tone of coverage regarding government and/or corporate actors’ reliability trustworthiness and dependability. This could be either general or with reference to the specific disaster. For example, the framing actor accusing another of a cover-up suggests lack of reliability/trust. Similarly, the framing actor expressing lack of confidence in existing procedures suggests lack of reliability/trust. Code as ‘neutral’ if the tone is apparent in the story and is fairly balanced (i.e., no positive or negative tone prevailing). 0 = Tone absent 1 = Unreliable 2 = Reliable 3 = Neutral 4 = Unclear

V15a. Framing actor Subject using the tone. Code as (0) if V16 is absent. 1 = Journalist 2 = Activist 3 = Local government 4 = Provincial/state government 5 = Federal/national government 6 = Corporation 7 = Expert 8 = Other 9 = Unclear

V16. Tone: information Tone of coverage regarding the provision of information on the disaster. provision Refers to specific actors’ willingness (or lack of it) to disclose information on the disaster. For example, code as ‘obscure’ if an actor argues the information should not be public. Code as ‘neutral’ if the tone is apparent in the story and is fairly balanced (i.e., no positive or negative tone prevailing). 0 = Tone absent 1 = Obscure 2 = Informative 3 = Neutral 4 = Unclear

V16a. Framing actor Subject using the tone. Code as (0) if V17 is absent. 1 = Journalist 2 = Activist

195

3 = Local government 4 = Provincial/state government 5 = Federal/national government 6 = Corporation 7 = Expert 8 = Other 9 = Unclear

V17. Tone: uncertainty Tone of coverage regarding the certainty/uncertainty surrounding the disaster. The uncertain tone can be conveyed through open acknowledgment of uncertainty (e.g., through explicitly referring to the uncertain nature of the disaster, whether in terms of causes or damages or other aspects) or through specific words such as ‘potentially’, ‘probably’, ‘likely’, etc. Only code the tone as ‘certain’ if certainty is explicitly stated (e.g., through the use of words such as ‘certain’, ‘sure’, etc., or through expressing high confidence in disaster causes, damages, etc.). Code as ‘neutral’ if the story/framing actor presents both certain and uncertain aspects of the disaster. 0 = Tone absent 1 = Uncertain 2 = Certain 3 = Neutral 4 = Unclear

V17a. Framing actor Subject using the tone. Code as (0) if V18 is absent. 1 = Journalist 2 = Activist 3 = Local government 4 = Provincial/state government 5 = Federal/national government 6 = Corporation 7 = Expert 8 = Other 9 = Unclear

V18. Tone: relatability Tone of coverage regarding the relatability of the disaster to the target audience, including the affected population or the public in general. Relatable tone can be conveyed through using comparable examples or human interest frames or referring to similar disasters potentially occurring in the future. Code as ‘neutral’ if the story/framing actor presents both relatable and unrelatable aspects of the disaster. 0 = Tone absent 1 = Relatable 2 = Unrelatable 3 = Neutral/balanced 4 = Unclear

196

V18a. Framing actor Subject using the tone. Code as (0) if V19 is absent. 1 = Journalist 2 = Activist 3 = Local government 4 = Provincial/state government 5 = Federal/national government 6 = Corporation 7 = Expert 8 = Other 9 = Unclear

197

Appendix 4

Figure 10. LDA topic model for the Mount Polley disaster (10 topics).

198

199

Figure 11. LDA topic model for the Deepwater Horizon spill (10 topics).

200

201

Figure 12. LDA topic model for the Fukushima disaster (Germany) (10 topics).

202

Appendix 5

Tone variables: 1= response, 2= preparedness, 3=dependability, 4=information, 5=uncertainty, 6=relatability

Figure 13. Tone and framing actors: journalist, Mount Polley.

60.0

50.0

40.0

Negative 30.0 Positive

Percentage Neutral 20.0 Unclear

10.0

0.0 1 2 3 4 5 6 Tone variables

Figure 14. Tone and framing actors: activist, Mount Polley.

14.0

12.0

10.0

8.0 Negative

6.0 Positive Percentage Neutral 4.0 Unclear

2.0

0.0 1 2 3 4 5 6 Tone variables

203

Figure 15. Tone and framing actors: local government, Mount Polley.

18.0

16.0

14.0

12.0

10.0 Negative

8.0 Positive Percentage 6.0 Neutral Unclear 4.0

2.0

0.0 1 2 3 4 5 6 Tone variables

Figure 16. Tone and framing actors: provincial government, Mount Polley.

30.0

25.0

20.0

Negative 15.0 Positive

Percentage Neutral 10.0 Unclear

5.0

0.0 1 2 3 4 5 6 Tone variables

204

Figure 17. Tone and framing actors: federal government, Mount Polley.

1.8

1.6

1.4

1.2

1.0 Negative

0.8 Positive

Percentage Neutral 0.6 Unclear 0.4

0.2

0.0 1 2 3 4 5 6 Tone variables

Figure 18 Tone and framing actors: company, Mount Polley.

4.5

4.0

3.5

3.0

2.5 Negative

2.0 Positive

Percentage Neutral 1.5 Unclear 1.0

0.5

0.0 1 2 3 4 5 6 Tone variables

205

Figure 19. Tone and framing actors: expert, Mount Polley.

18.0

16.0

14.0

12.0

10.0 Negative

8.0 Positive Percentage 6.0 Neutral Unclear 4.0

2.0

0.0 1 2 3 4 5 6 Tone variables

Figure 20. Tone and framing actors: other, Mount Polley.

16.0

14.0

12.0

10.0 Negative 8.0 Positive

Percentage 6.0 Neutral

4.0 Unclear

2.0

0.0 1 2 3 4 5 6 Tone variables

206

Figure 21. Tone and framing actors: journalist, Deepwater Horizon.

120.0

100.0

80.0

Negative 60.0 Positive

Percentage Neutral 40.0 Unclear

20.0

0.0 1 2 3 4 5 6 Tone variables

Figure 22. Tone and framing actors: activist, Deepwater Horizon.

7.0

6.0

5.0

4.0 Negative Positive 3.0 Percentage Neutral

2.0 Unclear

1.0

0.0 1 2 3 4 5 6 Tone variables

207

Figure 23. Tone and framing actors: federal government, Deepwater Horizon.

25.0

20.0

15.0 Negative Positive

Percentage 10.0 Neutral Unclear 5.0

0.0 1 2 3 4 5 6 Tone variables

Figure 24. Tone and framing actors: company, Deepwater Horizon.

35.0

30.0

25.0

20.0 Negative Positive 15.0 Percentage Neutral 10.0 Unclear

5.0

0.0 1 2 3 4 5 6 Tone variables

208

Figure 25. Tone and framing actors: expert, Deepwater Horizon.

20.0 18.0 16.0 14.0 12.0 Negative 10.0 Positive 8.0 Percentage Neutral 6.0 Unclear 4.0 2.0 0.0 1 2 3 4 5 6 Tone variables

Figure 26. Tone and framing actors: other, Deepwater Horizon.

25.0

20.0

15.0 Negative Positive

Percentage 10.0 Neutral Unclear 5.0

0.0 1 2 3 4 5 6 Tone variables

209

Figure 27. Tone and framing actors: journalist, Fukushima.

Negative 30 Positive

Percentage Neutral 20 Unclear

0 1 2 3 4 5 6 Tone variables

Figure 28. Tone and framing actors: activist, Fukushima.

3 Negative

2 Positive

Percentage Neutral 2 Unclear 1

0 1 2 3 4 5 6 Tone variables

210

Figure 29. Tone and framing actors: federal government, Fukushima.

10 Negative

8 Positive

Percentage Neutral 6 Unclear 4

0 1 2 3 4 5 6 Tone variables

Figure 30. Tone and framing actors: company, Fukushima.

15 Negative Positive

Percentage 10 Neutral Unclear 5

0 1 2 3 4 5 6 Tone variables

211

Figure 31. Tone and framing actors: expert, Fukushima.

8 Negative Positive 6 Percentage Neutral 4 Unclear

0 1 2 3 4 5 6 Tone variables

Figure 32. Tone and framing actors: other, Fukushima.

10 Negative 8 Positive

Percentage 6 Neutral

4 Unclear

0 1 2 3 4 5 6 Tone variables

212

Appendix 6

Experimental treatments.

213

214

215

216

217

Table 40. Factor loadings, principal components analysis.

Emotion Factor 1 Factor 2 Frustrated 0.7902 0.0221 Determined 0.0768 0.8293 Enthusiastic 0.0632 0.8381 Distressed 0.8950 0.0492 Alert 0.0123 0.6717 Afraid 0.8228 0.1356 Nervous 0.8131 0.0821 Scared 0.8225 0.1460 Upset 0.8292 0.0423 Angry 0.8039 0.0736 Anxious 0.7940 0.1143 Excited 0.8950 0.0492 Inspired 0.1421 0.8286 p(variance explained) 0.4800 0.2000

218

Debriefing Form for Deception Studies

During the experiment, you were asked to read a news article. The news article from Reuters describes a recent pipeline leak in central California that did not actually occur. In fact, the article was based on a real news story from CNN about a pipeline leak near Santa Barbara California in 2015. This article was shortened and changed to depict a recent event in central California. It was also altered to include references to uncertainty and human health impacts, which is the focus of our research.

These alterations were necessary to understand how different post-disaster conditions affect public willingness to participate in political action, and specifically nonviolent protest such as peaceful demonstrations, petitions or boycotts. It was necessary to make the event novel to understand how public participation emerges from certain framing of disasters. If we had selected an event from the past, it was unlikely that any willingness to participate in political action would have emerged, and, as a result, we would not have been able to test our theories.

We also redesigned the article to make it appear to be from the Reuters newswire service rather than from CNN. This change was essential so that participants would accept the information as credible. Citizens engage more with information when it is from sources they trust. Non-political newswire services are likely to be trusted by a wider range of Americans of different values and beliefs than a cable news network.

We have included links to the original article at the bottom of this page, for your reference.

Because you were deceived, you now have the right to refuse to allow your responses to be used and to ask that they be destroyed immediately. If you do so, there is no penalty. You will still receive full payment for the experiment.

_____ I give permission for my survey responses to be used in the analysis for this experiment.

_____ I do NOT give my permission for my survey responses to be used in the analysis for this experiment. Please withdraw them from the study and destroy them immediately.

News Story

Wildlife, pristine beaches focus of 'aggressive' oil spill cleanup, https://www.cnn.com/2015/05/20/us/california-oil-spill/index.html.

219