A “normalizing” re-alignment of voting behaviour in East Germany?

I am currently working on a book chapter about “Sozialstruktur und Wahlverhalten in Ost- und Westdeutschland – Konvergenz, Divergenz oder Persistenz?” (Social Structure and Voting Behaviour in East and West Germany – Convergence, Divergence, or Persistence?). In this context I found the remarkable result shown in the following figure.

../../../_images/mpred-cducsu.svg

Predictive margins of CDU/CSU support (vote intentions and reported votes), conditional on social class and time (controlled for church-attendance and election-specific random-effects.) The predictive margins were constructed with the help of the mpred package package, based on a random-effects multinomial baseline logit model fitted with the mclogit package package.

The results of the first Bundestag election after the German Unification in 1990 caught many by surprise: Firstly, it meant a massive electoral success of the Christian Democrats in the new Bundesländer although it had been widely expectet that the social democrats should gain most of the new votes of East Germans. Furthermore, the Christian Democrats were particularly strong in the relatively industrialised southern states of the East, most notably Saxonoy (Sachsen). Ironically, it appeared that the CDU had become the new “labour party” of the East. The data of electoral studies conducted since 1994 show, there indeed existed such a pattern, which however reversed over the course of the following elections (unfortunately, there is no generally available data on East German electoral behaviour in 1990).