Do Voters Join Unions or Do Unions Encourage Voting? : Evidence from the US Using a Latent Factor Potential Outcomes Model

Abstract

What is the causal effect of group membership on voter turnout? Extensive research on voting has documented a positive association between membership in groups like unions and participation in elections. But it is widely recognized that existing studies provide limited evidence on whether this reflects a causal relationship or is due to self-selection into union membership. We address this question using panel data from the US, the National Longitudinal Survey of Youth, which allows us to exploit credible strategies to deal with unobserved confounders. Building on recent advances in econometrics, we use a potential outcomes model with matching on both observable and unobservable individual characteristics. Unobservable characteristics are captured using a latent factor structure, which allows for proxies measured with error. We implement the model in a Bayesian framework. We find a clear average treatment effect of union membership. All else equal, union member are 10 percentage points more likely to turn-out on election day. We also find evidence of non-random selection into union membership.