Managing Militias: Recruitment, Discipline, and Governance among ...

[Pages:232]Managing Militias: Recruitment, Discipline, and Governance among Counterinsurgent Militias in Sierra Leone

Jonathan Filip Forney Charlottesville, Virginia

Bachelor of Arts in Political Science, University of California, Los Angeles, 2006 Master of Arts in Foreign Affairs, University of Virginia, 2008

A Dissertation presented to the Graduate Faculty of the University of Virginia in Candidacy for the Degree of Doctor of Philosophy

Department of Politics

University of Virginia July, 2014

c Jonathan Filip Forney 2014

Managing Militias: Recruitment, Discipline, and Governance among Counterinsurgent Militias in Sierra Leone

by Jonathan Forney

Abstract Why do non-state armed organizations change over time? Why do some armed groups retain high levels of discipline and cooperative relationships with civilians, while other groups degenerate into racketeering and rogue banditry? This dissertation uses micro-comparative evidence to identify the mechanisms of organizational durability and change within three counter-insurgent militias that operated in Sierra Leone from 1991 to 2002. Newly gathered oral history and survey data from interviews with over 150 former militia members reveal that militia recruitment processes are essential in managing community-level conflicts, and are thus highly predictive of the long-term trajectories of armed organizations that draw recruits from local communities. Critical questions about how militia members are recruited have profound consequences ? manifested in the varying levels of success of attempts to monitor and control existing militia members, and to extract key resources from civilian populations. My fine-grained examination of civil militias in Sierra Leone suggests important revisions to, and extensions of, existing theories of recruitment in armed groups and theories of civilian victimization during civil wars.

ii

For my American family, Barbara Forney and Peter Forney For my Sierra Leonean family, Etta Kamara, Bakarr Koroma, Sayed, and Waheb

iii

TABLE OF CONTENTS

LIST OF FIGURES

vi

LIST OF TABLES

vii

ACKNOWLEDGMENTS

ix

1 Introduction: Creating and Controlling Ad-hoc Armed Organizations

1

The Recruitment Problem: Finding Trustworthy Fighters . . . . . . . . . . . . . . 4

Scope Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

An Overview of the Dissertation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

2 A Theory of Recruitment and Organizational Change

18

Adverse Selection Problems: Finding Trustworthy Fighters . . . . . . . . . . . . . 21

The Effects of Recruit Quality on Discipline and Relationships with Civilians . . . 34

Research Design and Case Selection . . . . . . . . . . . . . . . . . . . . . . . . . . 38

3 Managing Militia Recruitment

44

Emergency, the Emergence of Militias, and Chiefly Leadership . . . . . . . . . . . 49

The Recruiters' Perspective: Who Could Chiefs Trust with a Gun? . . . . . . . . 53

Material Resources, Incentives, and the Supply of Volunteers . . . . . . . . . . . . 59

Using Information Networks to Find Loyal Fighters . . . . . . . . . . . . . . . . . 68

Explaining Convergence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77

Network Fragility and the Foreshadowing of Divergence . . . . . . . . . . . . . . . 79

iv

4 Crisis and Change: Recruiting the Wrong People

82

National-Level Crisis and Change: Better Organized, but Worse Off . . . . . . . . 86

Death, Displacement, Intrigue, and Divergence . . . . . . . . . . . . . . . . . . . . 90

Northern CDF Recruitment: Screening with Disrupted Networks . . . . . . . . . 92

Recruitment of Displaced Donsos . . . . . . . . . . . . . . . . . . . . . . . . . . . 98

Kamajor Recruitment and Rogue Initiators in the Deep South . . . . . . . . . . . 108

Summary and Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112

Appendix: Generating Sampling Weights to Adjust for RDS . . . . . . . . . . . . 114

5 Predicting Recruit Quality: Statistical Tests

116

Understanding Recruitment Outcomes . . . . . . . . . . . . . . . . . . . . . . . . 117

Data and Research Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121

Predicting Recruit Quality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135

Alternative Explanations: Maturation and Endogeneity . . . . . . . . . . . . . . . 142

Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147

Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148

6 Consequences of Recruitment: Discipline, Drugs, and Opportunistic Vic-

timization of Civilians

153

Understanding (In)discipline within Civil Militias . . . . . . . . . . . . . . . . . . 159

Understanding Civilian Victimization in Civil Wars . . . . . . . . . . . . . . . . . 169

Losing Control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183

Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 188

Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 190

7 Conclusion: How Things Fell Apart

195

Implications for the Study of State "Failure" and Informal Armed Groups . . . . . 196

Shortcomings of My Analysis and Avenues for Further Research . . . . . . . . . . 199

Beyond Civil Militias in Sierra Leone . . . . . . . . . . . . . . . . . . . . . . . . . 201

Bibliography

206

v

LIST OF FIGURES

3.1 Recruitment Map . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61 4.1 The Proportion of Sampled Militia Members Recruited by Year . . . . . . . 84 5.1 Model 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136 5.2 Model 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137 5.3 Model 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 137 5.4 Year Joined as a Predictor of Quality . . . . . . . . . . . . . . . . . . . . . . 142 6.1 Model 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164 6.2 Model 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165 6.3 Model 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165 6.4 Model 4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166 6.5 Instances of Civilian Victimization by District-Year . . . . . . . . . . . . . . 170 6.6 Relationship between Recruitment Selectivity and Civilian Victimization by

Year . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176

vi

LIST OF TABLES

2.1 The Selectivity of Screening Processes as a Function of Time and Information Networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

2.2 Summary of Case Study Variation . . . . . . . . . . . . . . . . . . . . . . . . 42

5.1 Recruitment Selectivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120 5.2 Homophily Estimates by Prior Firearms Experience . . . . . . . . . . . . . . 128 5.3 Homophily Estimates by Age Grouping . . . . . . . . . . . . . . . . . . . . . 128 5.4 Homophily Estimates by Ethnic Group . . . . . . . . . . . . . . . . . . . . . 128 5.5 Recruitment Selectivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134 5.6 Summary Statistics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135 5.7 OLS Regressions Predicting Recruit Quality - Binary Selectivity Variable . . 140 5.8 Regression Results Comparing Different Measures of Recruit Quality . . . . 141 5.9 Regressions Predicting Recruit Quality - Controlling for Population Degradation145 5.10 OLS Results: Determinants of Recruit Quality . . . . . . . . . . . . . . . . . 149 5.11 Post-Estimation Summary of Survey Design Effects for Model 2 (Above) . . 149 5.12 Ordered Logit Results: Determinants of Recruit Quality . . . . . . . . . . . 150 5.13 OLS Results: Recruit Quality, Omitting Optimal Screening . . . . . . . . . 151 5.14 Selectivity Scores by Region (District) and Year . . . . . . . . . . . . . . . . 152

6.1 Instances of Victimization of Civilians, by Type . . . . . . . . . . . . . . . . 175 6.2 Negative Binomial Regressions Predicting Civilian Victimization . . . . . . . 180 6.3 Summary Statistics for RDS Dataset ? Predicting Drug Use . . . . . . . . . 190 6.4 Summary Statistics for TRC Dataset ? Predicting Civilian Victimization . . 191 6.5 Logit Models Predicting Drug Use . . . . . . . . . . . . . . . . . . . . . . . . 192

vii

 ................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download