The centrality of checkpoints for civilians during conflict

Abstract

Analyses of the violence–displacement relationship motivate a disaggregation of violence into violence in residential areas (home violence) and violence along migration routes (road violence). These two forms of violence have opposite effects upon displacement, with home violence increasing displacement and road violence decreasing displacement. Yet, conflict scholarship would benefit from additional consideration of the specific activities that constitute road violence. This consideration reveals that road violence primarily involves displacement-deterring violence at checkpoints. This motivates the question: How does checkpoint violence deter displacement? Using interviews with Syrian and Somali refugees, as well as daily data on violence and displacement in Somalia, this paper argues that state and non-state armed groups create and foster uncertainty at checkpoints. Then, they use dramatic violence that is tangible to civilians through their ability to physically see, hear, and smell evidence of it as propaganda. As propaganda, the deterrent effect of checkpoint violence is amplified beyond isolated violent events. More broadly, results from quantitative analysis of the daily data on violence and displacement in Somalia indicate that uncertainty about road violence (road uncertainty) and uncertainty about home violence (home uncertainty) amplify the effects of road violence and home violence, respectively.

Acknowledgments

This paper has benefitted from the feedback of many people. Karen Rasler, William Thompson, and Lauren MacLean have all provided valuable comments and advising. Graduate students in the Department of Political Science at Indiana University Bloomington have provided comments at multiple stages of the project. Earlier versions of this research were also presented at the Ostrom Workshop at Indiana University, the 2014 Midwest Political Science Association Annual Convention, and a Spring 2015 session of the Conflict Consortium Virtual Workshop. Finally, I have received additional feedback from Thomas Zeitzoff and Tijen Demirel-Pegg. Financial assistance came from the Ostrom Workshop and a Foreign Language and Area Studies (FLAS) award from the Indiana University Center for the Study of the Middle East. All errors and omissions are my own.

Notes

1. Additional details about the interviews with Syrians and Somalis are contained in the Appendix 1.

2. Later events, such as the attack on the Westgate shopping mall in Nairobi, demonstrated that Al Shabaab may not have been as weak as previously thought. However, it is clear that Al Shabaab activity had substantially decreased from earlier levels.

3. ACF and PACF graphs are included in the Appendix. In addition, the results of AIC and BIC diagnostic tests are in the Appendix.

4. Weekly counts from 2007 were converted into daily counts by dividing each weekly count by 7. All data from 2008 were received by the author as daily data.

5. UNHCR has 48 total partners and at least two staff members per location in Somalia.

6. Personal communication with staff member at the UNHCR Field Office in Somalia on 9 December 2013.

7. Results are robust to using a band of 10 km around major roads as well.

8. This data-set relies upon news and NGO reports of violent events for its reporting of violent events and the characteristics of those events. It thereby suffers from the same problems faced by all event data-sets that rely upon news reports (Davenport 2010, Dulic 2010, Eck 2012).

9. Tests with the mean number of violent events, rather than the sum, produced similar results.

10. This measure initially leaves missing values on days when no violent events occur. Missing values are replaced with uncertainty measures that take an average of the most recent day with data before the missing value day and the next day with data. The assumption is that the characteristics of violent events are what influence changes in uncertainty about violence, so it is logical to estimate uncertainty for a given day based on the uncertainty of the preceding and subsequent days.

11. Technically, the AMISOM intervention began in 2007, overlapping with the Ethiopian intervention. However, AMISOM did not have a significant role, beyond securing the Mogadishu airport, until Ethiopian troops left in January 2009.