Many studies in the African politics literature estimate the correlation between ethnicity and vote choice using reported vote intentions from public opinion surveys. Yet, though we know questions related to ethnicity are sensitive in other contexts, there has been little investigation into whether or how survey conditions affect African respondents' willingness to report a preference for candidates of their own ethnicities. Using a stated choice experiment to identify the characteristics that Ugandan voters value in presidential candidates, I show that ethnic voting is indeed sensitive: Respondents are less likely to report a preference for coethnic politicians when they report their preference publicly or when they have been made conscious of the ethnic connotations of their choice. Both public exposure and priming have larger effects when respondents interact with non-coethnics, but this is not because respondents are more willing to report ethnic preferences to members of their own group. Rather, it is because those interacting with non-coethnics hold stronger implicit ethnic preferences in the first place. These results echo a series of findings from the United States that the presence of outgroup members activates ingroup biases, even as respondents censor the preferences they report to others and to themselves. The study indicates that gathering unbiased electoral preferences from African survey respondents will require granting respondents privacy and avoiding priming, and that the researchers should record and control for the ethnicity of observers.
All Science Journal Classification (ASJC) codes
- Sociology and Political Science
- Social Sciences(all)
- History and Philosophy of Science