Party Identification Measures in the Anglo-American Democracies: A National Survey Experiment

This paper stakes a prima facie claim that cross-national variation in party identification, at least among the Anglo-American democracies, is largely a measurement artifact, a product of the uniqueness of the American National Election Studies (NES) in prompting respondents directly for nonpartisanship. The heart of the analysis is a CATI experiment in a Canadian national survey. Respondents were randomly administered one of three variants of the party identification item. Respondents were also asked how they voted in the 1984 election and were randomly assigned to vote-before and voteafter treatments. When prompted for nonpartisanship, Canadians volunteer such response at roughly the same rate that U.S. respondents do. Response to the traditional Canadian item, which gives no nonpartisan prompt, is highly labile in the face of change in question order. Response to the alternative versions that prompt nonpartisanship is not affected at all by question order.