Abstract

In order to better understand the individual-level motives for ticket-splitting, Taiwan's Election and Democratization Study has since 2001 included a question aimed at measuring respondents’ preferences for checks and balances. We argue that this set of questions, designed to measure a combination of Fiorina's policy-balancing hypothesis and Ladd's cognitive Madisonianism, is inconsistent with principles of survey methodology and thus produces data that are suboptimal. Following a method developed by Carsey and Layman, we propose an alternative concept, the policy-balancing index derived from the perceived ideological distance between respondent and political parties, which both avoids methodological violations and provides us with a more precise concept to work with. We test the index and find it to be a significant determinant of ticket-splitting behavior