Does ticket-splitting decrease the accuracy of recalled previous voting? Evidence from three German panel surveys

Abstract

This paper addresses the role of ticket-splitting in decreasing the accuracy of recalled previous voting. It suggests that ticket-splitting makes recalling vote choice a more difficult task and thus increases recall error. Using data from three German panel surveys conducted from 1998 to 2008, the paper demonstrates that ticket-splitting increases recall error by a considerable margin, even after controlling for behavioral and attitudinal predictors. This finding suggests that recall accuracy is not a stable voter characteristic. It also lends credence to the notion that the increase in ticket-splitting in recent German elections decreased the accuracy of recalled previous voting. Finally, the evidence suggests that electoral systems affect the validity of survey answers concerning previous vote choice.