Selection and Response Bias in Protest Surveys

Abstract

Popular opinion suggests protest has become ubiquitous in advanced postindustrial democracies. In order to assess this perspective scholars have increasingly relied on so-called protest surveys to gauge protesters' attitudes and characteristics while in the act of protesting. This growing body of literature has neglected the design of protest surveys and has yet to systematically test the reliability of the evidence they produce. This study attempts to make a contribution on both fronts. After reviewing the available studies relying on protest surveys, it proposes a standardized method to sample respondents in moving crowds. The sampling procedure is tested in 22 demonstrations across a variety of issues and countries. Based on a series of field experiments, the paper puts forward a specific protest survey fieldwork method by which selection bias and response bias can be reduced to a considerable degree.