UNIVERSITY OF CALIFORNIA, SAN DIEGO Essays on Protest Mobilization in Authoritarian Regimes A Dissertation submitted in partial satisfaction of the requirements for the degree Doctor of Philosophy in Political Science by Zachary Caleb Steinert-Threlkeld Committee in charge: Professor Emile M. Hafner-Burton, Chair Professor James H. Fowler, Co-Chair Professor J. Lawrence Broz Professor David A. Lake Professor Barbara F. Walter 2016 Copyright Zachary Caleb Steinert-Threlkeld, 2016 All rights reserved. The Dissertation of Zachary Caleb Steinert-Threlkeld is approved, and it is acceptable in quality and form for publication on microfilm and electronically: Co-Chair Chair University of California, San Diego 2016 iii DEDICATION To my parents, Kayte and Tom. iv EPIGRAPH Disobedience is the true foundation of liberty. | Henry David Thoreau v TABLE OF CONTENTS Signature Page . iii Dedication . iv Epigraph . .v Table of Contents . vi List of Figures . ix List of Tables . .x Acknowledgements . xi Vita......................................... xvi Abstract of the Dissertation . xvii Chapter 1 Spontaneous Collective Action: Peripheral Mobilization During the Arab Spring . .1 1.1 Abstract . .2 1.2 Introduction . .2 1.3 Theory . .5 1.3.1 Scope Conditions . 14 1.4 Data . 16 1.5 Measures . 22 1.5.1 Coordination . 22 1.5.2 Core Coordination . 26 1.6 Model . 30 1.7 Results . 34 1.7.1 Verification . 37 1.8 Exogenous Identification of the Core in Egypt . 49 1.9 Discussion . 52 1.10 Supplementary Materials . 58 1.10.1 Twitter Data . 58 1.10.2 Construction of Coordination Measure . 58 1.10.3 Alternate Measures of Elites . 65 1.10.4 Exploring Upper Tail of Core Distribution . 67 1.10.5 Potential Model Misspecification . 73 1.10.6 Placebo Tests . 78 1.10.7 Events Data . 81 vi 1.10.8 Deeper Analysis of Egypt, Bahrain, Morocco, and Qatar . 90 1.10.9 Shared and Unique Words for Hashtags . 94 1.10.10 Topic Model Detail . 97 1.11 Acknowledgements . 100 Chapter 2 Activism and Protest in Authoritarian Regimes . 101 2.1 Abstract . 102 2.2 Introduction . 102 2.3 Theory . 104 2.4 Research Design . 119 2.5 Results . 127 2.6 Activists and Policy Negotiation . 138 2.7 Conclusion . 141 2.8 Supplementary Material . 145 2.8.1 Detail on Topic Models . 145 2.8.2 Mobility . 148 2.8.3 Hashtag Topics by Social Movement . 149 2.8.4 Blackout . 156 2.9 Acknowledgements . 162 Chapter 3 Longitudinal Network Analysis with Incomplete Data . 163 3.1 Abstract . 164 3.2 Introduction . 165 3.3 Two Measures for Network Analysis . 167 3.3.1 Inference of Edge Formation . 167 3.3.2 Centrality Measurement with Incomplete Data . 171 3.4 Network Data . 175 3.4.1 Twitter . 175 3.4.2 Network Data from Twitter . 178 3.5 Activism During the Arab Spring . 181 3.5.1 Background . 181 3.5.2 Acquiring Activism Data . 182 3.6 Application . 184 3.6.1 Reconstructing Daily Network Change . 184 3.6.2 Daily Changes in Influence . 192 3.7 Conclusion . 195 3.8 Acknowledgements . 199 Chapter 4 Twitter as Data . 200 4.1 Abstract . 201 4.2 Introduction . 201 4.3 Why Twitter? . 202 4.4 Acquiring Data from Twitter . 205 vii 4.4.1 Purchase . 205 4.4.2 Collaborate . 207 4.4.3 Acquire on Your Own . 209 4.4.4 Access Constraints and Replication . 214 4.4.5 A Note on Programming . 217 4.5 Types of Questions . 220 4.5.1 Networks . 220 4.5.2 Text . 222 4.5.3 Spatial . 224 4.6 Conclusion . 225 4.7 Acknowledgements . 228 Bibliography . 229 viii LIST OF FIGURES Figure 1.1: Protests per Million Inhabitants . 18 Figure 1.2: Marginal Effects of Peripheral and Core Coordination . 36 Figure 1.3: Verifying Operationalization of Coordination . 38 Figure 1.4: Change in Effect Size as Function of Core Threshold . 41 Figure 1.5: ICEWS Correlates with Handcoded Data . 48 Figure 1.6: Blackout Influentials and Protest . 52 Figure 1.7: PDF ! CDF ! Gini . 60 Figure 1.8: Least Coordination: Bahrain on 02.08.2011 . 61 Figure 1.9: Most Coordination: Egypt on 02.02.2011 . 62 Figure 1.10: Egypt on 01.24.2011 . 62 Figure 1.11: Core Coordination Variable . 63 Figure 1.12: Core Coordination Variable . 64 Figure 1.13: Daily Average Tweet Production by Popularity Threshold . 65 Figure 1.14: Elite Follower Ratio by Popularity Threshold . 66 Figure 1.15: Coordinationi;t Peaks with 1-Day Lag . 80 Figure 1.16: More Protestors Correlate with More GDELT Protests . 83 Figure 1.17: ICEWS Protest Recording, 4 Countries . 87 Figure 1.18: Hashtag Intensity, Word Varies Based on Events . 92 Figure 1.19: Hashtag Use Varies by Coordination . 93 Figure 2.1: Mobility in Egypt, Bahrain . 149 Figure 2.2: Egypt - Activists' and Non-Activists' Use of Common Hashtags . 150 Figure 2.3: April 6th - Activists' and Non-Activists' Use of Unique Hashtags . 151 Figure 2.4: NoMilTrials - Activists' and Non-Activists' Use of Unique Hashtags 152 Figure 2.5: Anti-Sexual Harassment - Activists' and Non-Activists' Use of Unique Hashtags . 153 Figure 2.6: Bahrain - Activists' and Non-Activists' Use of Common Hashtags 154 Figure 2.7: Bahrain - Activists' and Non-Activists' Use of Unique Hashtags . 155 Figure 2.8: Activists' and Non-Activists' Differential Twitter Activity . 157 Figure 2.9: Egypt - Differential Hashtag Use . 159 Figure 2.10: Bahrain - Differential Hashtag Use . 160 Figure 3.1: Pseudocode for Inferring Connection Date . 170 Figure 3.2: Degree Centrality = 5; NCC = 9 . 173 Figure 3.3: Verification Against Ground Truth Data . 187 Figure 3.5: Bahrain and Egypt Activists, 01.24.2011 . 189 Figure 3.6: Bahrain and Egypt Activists, 04.04.2011 . 190 Figure 3.7: Reconstructed Temporal Change in Influence . 194 ix LIST OF TABLES Table 1.1: Comparing Core Measure with Hand-coded Accounts* . 27 Table 1.2: Variable Correlation . 32 Table 1.3: Variables by Country . 33 Table 1.4: Peripheral Coordination and Protest . 35 Table 1.5: Robust to Operationalization of Core, Periphery . 40 Table 1.6: Core Threshold Descriptive Statistics Across Countries . 44 Table 1.7: Robust to Dependent Variable . 46 Table 1.8: Blackout Accounts do not Provide Coordination . 53 Table 1.9: Tweet Typology . 59 Table 1.10: Core Threshold Statistics by Country . 68 Table 1.11: Robust to Model Specification . 74 Table 1.12: Robust to Removing Countries . 77 Table 1.13: Coordination in the Hashtag Long Tail . 79 Table 1.14: ACLED Protests . 89 Table 1.15: Topic Models' Parameters . 99 Table 2.1: Descriptive Statistics from Sifter Data . 122 Table 2.2: Topic Models' Parameters . 126 Table 2.3: Main Results . 133 Table 2.4: Robustness Checks - Clustered SEs, Count IVs . 135 Table 2.5: Robustness Checks - Tweets from Phones . 137 Table 2.6: Tweets by Category . 147 Table 2.7: Topic Models' Parameters . ..