Political polarization is a growing concern in many countries. Are mass protests merely a sign of increasing cleavages, or do they polarize societies? In this paper, we estimate the impact of Nazi marches in 1932 Hamburg, using granular data from 622 voting precincts during 6 elections. We show propaganda can convince – but it does so the most in areas with high initial support. Importantly, marches can also backfire, repelling voters. Thus, political campaigning leads to polarization. These effects diffused through social networks, measured as contagion patterns across neighborhoods from the 1918 Spanish flu outbreak. The electoral effects of social spillovers are of similar importance as direct exposure, and grow over time.

Democracies increasingly use censorship to counter foreign propaganda, yet evidence on its consequences remains scarce. We exploit the European Union’s 2022 ban on Russia Today (RT) and Sputnik as a natural experiment, using a triple-difference design that compares users connected to the banned outlets against unconnected users in EU and non-EU countries. We analyze a daily panel of 677,780 tweets from 146,633 Twitter users in seven European countries. Pro-Russia output declines by 21.7% among connected EU users; in a difference-in-differences comparison with non-EU users, total pro-Russia output among EU users falls by 13.6%. Alternative suppliers do not fill the gap: neither their pro-Russia output nor the engagement they receive rises after the ban. Consistent with an agenda-setting role of the banned outlets, the share of EU users’ tweets covering the outlets’ daily top-five topics decreases by 17%. A survey experiment offers suggestive evidence that such censorship can come at a cost to the very norms it is meant to defend.

This paper examines how new identity markers emerge, spread, and define enduring social boundaries. I study Reformation Germany (1480–1806) using 4.9 million baptismal names from 2,210 towns to trace how families expressed religious affiliation through naming choices. Naming remained nearly identical across Catholics and Protestants for decades after 1517 but diverged sharply from 1580 onward, stabilizing by 1618. Church enforcement strengthened local identities, while social learning and conformity spread them across space. Using a staggered difference-in-differences design around Jesuit and Protestant college foundings—complemented by counterfactual placements and visitation records—I show a strong link between institutional presence and local alignment. Applying a DeGroot model of learning with spatial-lag specifications, I show that towns updated their naming toward locally shared beliefs and prevailing norms, producing lasting cultural and economic divides.