Research

How people shape each other's political affiliation is a central question in the social sciences—with broader implications for partisan sorting and polarization—but is difficult to answer because people select into peer groups. We study a natural experiment in which the Church of Jesus Christ of Latter-day Saints periodically redraws congregation assignment boundaries, reassigning members to new church peers. Linking records on every Utah registered voter from 2012–2021 to digitized congregation maps, we employ a spatial difference-in-discontinuities design that compares changes in party registration among voters who live near one another and start off in the same congregation but are assigned to different congregations. Voters conform to their new peers: Unaffiliated voters assigned to more Republican congregations become more likely to register Republican, and Republicans assigned to more Democratic congregations become more likely to leave the party. Reassigned boundaries show no relationship with prior registration changes, supporting the design’s identifying assumption. The effects are not explained by correlated peer characteristics such as race or by the partisanship of congregation leaders. Effects are strongest between demographically similar peers, the members most likely to become friends, suggesting that influence operates through personal relationships among co-congregants.