Quieted by Questions: The Unintended Consequences of Survey Interviews on Protest in Africa

Survey research has long been a cornerstone of comparative politics, yet little quantitative evidence exists regarding its aggregate political effects in developing countries. We address this gap by arguing that respondents may misperceive academic survey interviews as state surveillance, deterring them from protesting. Leveraging the random assignment of Afrobarometer interviews, the plausibly exogenous assignment of “official-like” interviewers, and a difference-in-differences design, we show that survey interviews have sizable effects on protest in Africa. When no respondent perceived the survey as government-sponsored, interviews increased the likelihood of protests by 538% relative to the pre-treatment average (an encouragement effect). In contrast, when all respondents perceived government sponsorship, survey interviews reduced the likelihood of protests by 458% relative to the pre-treatment average (a deterrence effect). These effects are primarily driven by surveys conducted in non-liberal democracies, whereas they are different or even opposite in other political regimes. Extensive robustness and mechanism checks provide further support. These findings imply unintended externalities of survey research on real-world politics.