ABSTRACT
Does the probability to join a political party, to become a party functionary, and to leave a party depend on individuals’ socioeconomic status? Political parties are central mediating actors between the population and the state; thus, it is reasonable to assume that unequal participation within parties fosters unequal political representation. However, due to data limitations no study has hitherto examined the social selectivity of the whole party membership cycle. We shed light on these issues by analysing original data from the German Party Membership Study 2017. We find that socially disadvantaged individuals are less likely to become and to stay party members and have a lower proclivity for holding political offices. These effects persist even after controlling for social-psychological variables and the general incentives for party membership. However, in line with recent findings on voter turnout we show that social selectivity is partly mediated by political efficacy.
Notes
1 Typically, this problem is lessened by contrasting a party's electorate to its membership.
2 These comparisons of frequency distributions are complemented by multivariate models. Their analyses, however, do not consider the group of non-members as a whole but differentiate between those who can imagine being a party member and those who cannot. In this context, only the SES-differences between party members and individuals being sure not to join a party are statistically significant.
3 The correlation between political participation and political efficacy is partly attributable to the effect of political participation on efficacy (Finkel, Citation1985). Is this reverse causality a problem for our argumentation? The fact that political self-confidence improves with political work is independent of how high an individual's SES is. Accordingly, it cannot be assumed that this mechanism would be responsible for a disappearing SES effect when we control for efficacy.
4 Those respondents from the telephone survey who had never been a member of a party had some problems assessing the costs of membership. For this reason, missing values were replaced by mean value imputations, with the average calculation and the value substitution being carried out separately for current and former party members as well as non-members.
5 According to the Hausman test, our outcome categories do not have the property of independence of irrelevant alternatives. Therefore, instead of using multinomial logistic analyses, we decided to estimate multinomial probit models which relax the assumption of independence of irrelevant alternatives.
6 The comparison between former members and non-members results logically from the difference between the coefficients of and . Although this contrast is irrelevant for our focus, we have included a corresponding table (A1) in the online appendix to the document with the complete results. It is available through the following Figshare DOI: https://doi.org/10.6084/m9.figshare.11357705.
7 We considered the positions of executive board members, chairs of the executive board and elected officials at district, state, federal and European level.
8 We weight the data to ensure that our results apply to all members of the six parties we surveyed. For one thing, we correct for the disproportionate sample (see section 4). For another, the members of the various parties are weighted relative to each other according to the number of members of their parties.