(With Susanne Baltes and Donald Green)
The concept of party ID has long been central to the study of voting behavior. In their influential early study, The American Voter, Campbell et al. (1960) define party ID as “an individual’s affective orientation to an important group-object in his environment” that often results in a long-term attachment to a political party. Although not immune to the impact of life experiences and political events, party ID is considered an essentially stable personal characteristic. Some authors interpret party ID as a“genuine form of social identification’’ (Green, Palmquist and Schickler 2002), akin to other categories of social identification such as class, sex and religion. To dramatize the hypothesized stability of partisan orientations over time, party ID is often referred to as the “unmoved mover” in political behavior—a stable trait that determines an individual’s response to current political events and subsequent voting and other political behavior. Indeed, the relative stability of party ID is an important prerequisite if the concept is to serve as an explanatory variable that is conceptually distinct from short-term forces such as vote intentions or current evaluations of incumbent performance and party platforms.
Rejecting the hypothesized stability and explanatory importance of party ID, critics both in the United States and in Germany have charged that individuals adjust their partisanship in the wake of such short-term forces. In the United States, scholars have argued that partisanship responds to (1) the evaluation of presidential performance in office (Brody & Rothenberg 1988, Fiorina 1981), (2) the relative attractiveness of political platforms (Franklin & Jackson 1983), and (3) affinity for presidential candidates (Page & Jones 1979). In Germany, skeptics of party ID bring forth two important criticisms. First is the well-known critique of measurement error in surveys, which according to Küchler (1995) precludes valid and reliable measurement of party ID. Second, critics argue that the commonly asked questions merely induce “pseudo-identification” (Gabriel 2000). These critiques have been countered with panel survey evidence both in Germany (Arzheimer and Schön 2005) and in the US (Green, Palmquist and Schickler 2002). Still, the stability of party ID remains controversial (Miller and Klobucar 2005; Brader and Tucker 2001; Bartle and Bellucci 2009).
We add to the ongoing debate by conducting a large-scale four-wave panel survey in Germany and, eventually, in other Western European countries. Our aim is to fill in methodological and conceptual gaps in existing work. Like any social psychological construct, party ID is a latent variable that must be measured indirectly by survey questions that assess the concept as closely as possible. Even careful design and implementation of survey measures, however, inevitably encounter problems of reliability and validity (Converse and Pierce 1985). Our study, therefore, makes three improvements on existing work. First, we employ a multi-measure approach to the assessment party ID, which to our knowledge has not yet been utilized in Germany. In contrast to psychometric approaches, which commonly rely on multi-item measurements, political scientists have seldom employed redundant measures of the same concept, possibly because of a lack of resources. By asking a series of questions relating to party ID, we will triangulate the latent variable and reduce measurement error. The questions we use have all been successfully tested in other settings but have yet to be employed in combination or in the context of a national probabilistic sample. Second, our design leverages the benefits of over-time measurement afforded by panel surveys. Going forward, this will further increase measurement accuracy, enable the estimation of time-trends, and allow for party ID measurement at the individual level. Third, we incorporate advances in survey wording, including social psychological dimensions of party ID. In addition to using the classic Michigan-style wording commonly used in Germany (Question 1), we employ questions drawn from the literature that tap (1) the social-psychological underpinnings of party ID (Greene 2002), (2) the evaluation party performance (Fiorina 1981), and (3) the evaluation of party policies (Franklin & Jackson 1983).