Identifying the True Party Identifiers

<jats:p> The paper reviews the evidence that the measure of party identification (or partisanship) routinely used in social surveys and national election studies is flawed due to the fact that it overestimates the number of party identifiers in any given sample. The results of a 12-month survey experiment are reported to show that an alternative measure of party identification, one that does not prompt respondents with the names of the political parties and allows respondents a clear opportunity to register a `non-identity', substantially reduces the proportion of reported party identifiers. </jats:p>