The insurrection at the United States Capitol on January 6, 2021, was the most dramatic contemporary manifestation of deep political polarization in the United States. Recent research shows that violent protests shape political behavior and attachments, but several questions remain unanswered. Using day-level panel data from a large sample of US social media users to track changes in the identities expressed in their Twitter biographies, we show that the Capitol insurrection caused a large-scale decrease in outward expressions of identification with the Republican Party and Donald Trump, with no indication of reidentification in the weeks that followed. This finding suggests that there are limits to party loyalty: a violent attack on democratic institutions sets boundaries on partisanship, even among avowed partisans. Furthermore, the finding that political violence can deflect copartisans carries the potential positive democratic implication that those who encourage or associate themselves with such violence pay a political cost.

Rising inequalities in rich countries have led to concerns that the economic ladder is getting harder to climb. Yet, research on trends in intergenerational income mobility finds conflicting results. To better understand this variation, we adopt a multiverse approach that estimates trends over 82,944 different definitions of income mobility, varying how and for whom income is measured. Our analysis draws on comprehensive register data for Swedish cohorts born 1958 to 1977 and their parents. We find that income mobility has declined, but for reasons neglected by previous research: improved gender equality in the labor market raises intergenerational persistence in women’s earnings and the household incomes of both men and women. Dominant theories that focus on childhood investments have blinded researchers to this development. Methodologically, we show how multiverse analysis can be used with abduction—inference to the best explanation—to improve theory-building in social science.