Measuring party identification: Britain, Canada, and the United States

The article proposes an empirically based reflection on how to measure party identification cross nationally using data from the 1997 Canadian Election Study, the 1997 British Election Study, and the 1996 American National Election Study. These studies included both traditional national questions and a new common one, which allows for an assessment of the effects of question wording on the distribution and correlates of party identification. We show that the distribution of party identification is strongly affected by question wording and that the relationship between party identification and variables such as party and leader ratings, and voting behavior does not quite conform to theoretical expectations. We point out problems in the wording of party identification questions and propose an alternative formulation.