Employment Status and Political Beliefs - the non-causal Tale. What Public Opinion Research can Learn from Propensity Score Matching

How employment status affects the way we think about politics and how we behave in the political realm is at least since the Marienthal study by Jahoda et al. a recurring topic in the study of political behavior and political beliefs. Lately, it has regained prominence due to David Rueda's labor market dualization theory, that draws on labor market policy (LMP) attitudes of different employment status groups as individual mechanism. For the existing studies, the topic is rather uncontroversial: unemployed people benefit from policies aimed at creating jobs (active LMPs) or supporting the unemployed in financial terms (passive LMPs). The same is true for people in insecure employment relations (e.g. temporary employment). After all, they expect to get unemployed sooner or later. So, due to self-interest reasons, employment status will affect the way you think politically. Empirical Results regarding LMP attitudes seem to support this reasoning. However, I believe that it's too hasty to infer a causal influence of employment status on political beliefs from that. I do so, due to theoretical as well as methodological deficiencies pertinent to the existing literature. On a theoretical level, the existing literature does not take into account sufficiently the impact of relatively stable political predispositions (party attachments, ideology, value orientations) on policy attitudes. Moreover, these predispositions are not randomly distributed. Starting from early childhood, these predispositions are shaped by (i.e. socialized dependent on) the socio-economic context, in which one is raised. But your socio-economic background does not only shape your political predispositions. It also influences your risk to get unemployed or an insecure employment. Thus, any association between employment status and political attitudes, that does not account for all (!) factors that shape labor market risks and socialization, is confounded (i.e. exists due to a common cause: socio-economic background). Confounders, that are typically overlooked in the previous literature are for example parents characteristics. Fortunately, Propensity Score Methods allow us to think and handle the problem of confounding properly. Propensity Score Matching is done in three steps. 1) Computing an assignment model, i.e. a logit model on the chance of being unemployed. 2) Matching on the Propensity Score. 3) Causal Effect Estimation In empirical terms I will use data from the European Social Survey to look at first at the relation between employment status and party attachments as well as left-right self-placement. I hope to show that these association are not causal, i.e. that they vanish after Propensity Score Matching. In a second step, I will do the same for LMP attitudes as well as for other welfare state attitudes. The idea that your socio-economic background shapes your basic political orientation is a rather general one. Furthermore, social background does not only influence labor market risks, but also income, vocational training, etc. Therefore, I think that Propensity Score Methods are beneficial for a variety of questions in research on political beliefs and political behavior - as a method, and as a way of thinking about causality in general.