How Social Desirability Response Bias May Lead to an Overestimation of Obama-Trump Voters

Abstract

Abstract In this research note, we argue that a segment of Trump voters may be motivated to demonstrate that they are not racially insensitive by saying they voted for Obama in 2012. As a result, what may appear as a substantial group of swing voters may be—in part—partisan voters who want to avoid perceptions of racial bias. Using data from a survey of a nationally representative nonprobability sample from YouGov, polling data from the 2016 election, and a survey experiment conducted with a convenience sample of American adults, we find that Obama-Trump voters score significantly higher on a social monitoring scale than others and that respondents are significantly more supportive of Trump when they are given the opportunity to mention who they voted for in 2012. The combination of results suggests that the number of Obama-Trump voters may be overestimated due to self-monitoring.