Why do people vote in large elections? Theoretical arguments to resolve this paradox of voting often emphasize individuals’ pro-social motivations, which make turnout decisions less sensitive to the pivot probability. Using Norwegian population-wide register data, we test this argument by leveraging population-size shocks from inter-municipal mobility and proxying pro-social motivations via individuals’ charitable donations. We find that increasing electorate size widens the turnout gap between more/less pro-social individuals, and that turnout of pro-social individuals responds less to population-size shocks. Simulated municipality-level pivot probabilities suggest the (expected) probability of influencing election outcomes as a driving force behind these findings.