Do Random Response Questions Really Elicit Truthful Answers to Sensitive Questions? The Case of the Mississippi Personhood Initiative

Abstract

Random response questions have been used as a survey technique to elicit candid responses to sensitive questions for half a century, despite relatively little evidence as to their effectiveness in doing so. We reconsider surprising recent evidence that random response questions apparently were quite effective in eliciting candid retrospective responses about voting behaviour in a controversial ballot initiative in Mississippi. We argue that these findings likely are the result of a fortuitous respondent misunderstanding of an unorthodox formulation of a random response question that should not have encouraged candour had respondents understood its properties. Generalizable evidence on the effectiveness of random response questions remains elusive.

Keywords: reticence, random response questions, voting behavior, surveys

JEL Classification: C83, D72