| Abstract: | Direct survey measures of turnout often suffer from misreporting, particularly among non-voters. We investigate whether turnout misreporting in online surveys can be reduced by two new turnout question designs aimed at strengthening or buffering respondent's self-integrity against the perceived psychological threat of admitting non-voting. Drawing on evidence from survey experiments embedded in vote validated online surveys after the 2016 UK EU Referendum, we find that neither technique significantly improves turnout reporting accuracy. Our findings inform innovations in survey measurement of turnout and sensitive survey topics more generally. |
