Police, non-state actors, and political legitimacy in Central America

ABSTRACT

This paper explores the relationship between police presence and political legitimacy in countries with limited statehood. Using cross-sectional, national-based survey data in Central America’s northern triangle countries, this article measures the impact of the territorial police presence and other security-related groups on attitudes of support for the political system and democracy. The findings show that the perceived police presence at the community level is associated with higher political legitimacy in Guatemala, El Salvador, and Honduras. Such presence is also associated with elevated attitudes of support for democracy in El Salvador and Honduras. However, the results also indicate that other non-state actors can contribute to the support for the political system. The article discusses the implications of these findings in states with limited territorial reach.

KEYWORDS:

• Police
• political legitimacy
• gangs
• Central America

Acknowledgments

The author is grateful to David Holiday and Angélica Zamora for the support of this research, and to Jenny Fleming, Sebastian Roché, Jonathan Rosen, and the anonymous reviewers for comments on previous versions of this article.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 El Salvador is the smallest country in the continental Americas, with the highest population density, and better human development indicators than the neighbouring countries, except for Costa Rica (Vargas Cullell 2013).

2 I am grateful to LAPOP’s director, Elizabeth Zechmeister, and to IUDOP’s director, Jeannette Aguilar, for their support.

3 Information regarding the Latin American Public Opinion Project (LAPOP) and the methodology of the AmericasBarometer can be found at: http://sitemason.vanderbilt.edu/lapop/about_us.

4 I ran tests to detect specification errors in all models presented here. I used robust standard errors to address potential heteroskedasticity in some models. I also estimated the Variance Inflation Factor (VIF) statistics to detect potential problems of multicollinearity in the explanatory variables. The results indicated no significant multicollinearity issues among the variables used in the models (VIFs < 2).

Additional information

Funding

This paper results from a research project on policing in Central America, funded by the Open Society Foundations [Grant Numbers OR2015-22547 and OR2016-31214].