| Abstract: | Party identification (PI) is one of the central theoretical concepts in political sociology. Many scholars, however, argue that the lack of stability in observed PI casts serious doubts on its role as an “unmoved mover” ascribed to the concept by the Michigan school. The aim of this paper is twofold. First, we argue that perfect stability is not required by the Michigan approach. Rather, PI should be assumed to vary systematically in response to political and extra-political conditions. Second, we show that PI is a latent variable whose stability is massively underestimated if random measurement error is not taken into account. With a Latent Transition Analysis of the GSOEP, we can demonstrate that a) PI is highly stable amongst supporters of the major parties in both East and West Germany, b) that the stability of PI varies somewhat with the degree of political interest, and that c) stability has not declined since the mid-1980s. |
