Of Artifacts and Partisan Instability

In this article we examine the effects of measurement error on the apparent stability of party identification. As others have noted, party identification appears to be highly stable once measurement error is taken into account. This seems to be in conflict with recent models that claim to demonstrate the responsiveness of partisanship to short-term political forces. Using data from the 1980 National Election Study Major Panel, we demonstrate that when measurement error is included explicitly in models of political behavior, the effects of performance evaluations, candidate evaluations, issue proximities, or vote choices on party identification are negligible. Thus, it appears that the findings that show a significant impact of these short-term forces on partisanship represent artifacts of inadequate measurement model specification. Our results suggest that the outstanding characteristic of party identification is its persistence over time.