Practical and Institutional Barriers to Statistical Best Practices in Political Science

• Activity Number: 499
• Type: Topic Contributed
• Date/Time: Wednesday, August 3, 2011 : 10:30 AM to 12:20 PM
• Sponsor: Social Statistics Section
• Author(s): Amelia Hoover Green*+
• Companies: Yale University

Abstract

Political science research on wartime violence against civilians frequently relies on quantitative data. Yet it has seldom grappled with the measurement issues afflicting political violence data. This inattention to facts on the ground may lead to serious inferential errors. However, both practical and institutional barriers have hindered the adoption of statistical best practices in conflict studies. One such 'best practice' is multiple systems estimation (MSE). In this paper, I describe a multi-method study of violence against civilians during civil war in El Salvador, which employs MSE to correct raw data on violence. I review the project's substantive findings, in particular the estimated death toll and the asymmetry of violence between government and insurgents. Additionally, I describe traditional analytical strategies for this type of research question (linear regression or qualitative analysis), and how they differ from MSE-based strategies, both methodologically and substantively. Most importantly, I consider the analysis and academic reception of this study as an example of the practical and institutional barriers to statistical best practices in political science.