Research

We estimate the causal effect of peer partisan composition on party affiliation and voter turnout. Our setting is Utah, where members of the Church of Jesus Christ of Latter-day Saints are assigned to geographically defined congregations ("wards"). We exploit periodic ward boundary adjustments—made to accommodate changes in local church attendance—using a geographic difference-in-discontinuities design that compares changes in political outcomes among individuals who belonged to the same ward at baseline but were subsequently reassigned to wards with different partisan compositions. We find that peer composition affects party affiliation and voter turnout through distinct channels. For party affiliation, exposure to co-partisan peers increases the likelihood of adopting that party: being assigned to the more Republican side of a boundary raises an Independent's probability of registering Republican by about 1.3 percentage points. For voter turnout, we find effects consistent with negative mobilization: co-partisan exposure reduces turnout by about 1 percentage point, while cross-partisan exposure increases it by 1–2 percentage points. A placebo test using future boundaries and a simulated-boundary exercise both support the identifying assumption. Heterogeneity analysis suggests that party switching is driven by persuasion through close (same-gender) ties, while turnout effects operate through broader exposure to opposing views. Our results indicate that sustained interaction in a nonpolitical congregational setting has a small but robust causal effect on political identity and civic participation.