Nina Nikiforova

We study the communication strategies on Twitter-X of 367 political leaders in 21 countries, focusing on electoral competition between populists and non-populists. We measure polarization by the ease with which the leader can be classified as populist or not, conditional on their tweet. We find that political rhetoric becomes more polarized before and around election dates. This happens because, in pre-electoral quarters, opposite leaders are more likely to: (i) talk about different topics, and (ii) frame differently the same issues. Our results are consistent with competing politicians targeting different voters, rather than appealing to the same swing voters.

I propose a method to identify which ideological dimensions matter to voters. It applies a spatial voting framework to partial party rankings: a dimension rationalizes any ranking consistent with proximity voting along it. I take the method to the German Longitudinal Election Study, 2013–2022, which elicits respondents' three most-preferred parties, together with expert survey placements of parties on seven ideological dimensions (Left–Right, economy, culture and anti-elite, among others). In one dimension, Left–Right rationalizes 40–50% of reported rankings in every year, more than any dimension considered and more than a random one would; from 2017 onward, no conceivable dimension rationalizes a larger share. In two dimensions, the economy–culture pair performs best. Both models explain less over time, while the anti-elite dimension gains explanatory power, rising by about 2 percentage points per year and matching the economic dimension by 2022; this rise is driven by supporters of mainstream parties. These patterns are consistent with a shift in political conflict from the traditional Left–Right dimension toward an anti-elite, or populist, dimension.

Do political preferences shape where students choose to study? We investigate this question in France, combining administrative data on the universe of Parcoursup applicants with a conjoint survey experiment conducted with 1,040 final-year high school students. In the administrative data, students are more likely to apply to universities whose campus political climate aligns with the political leaning of their ZIP code, controlling for student, university, and major fixed effects, and the same applicant tailors the political content of her statements of purpose to the leaning of each institution. In the conjoint experiment, where alignment is randomly assigned, campus political climate carries a weight comparable to that of distance: students are willing to travel approximately one additional hour to attend a politically aligned institution, and 72% of surveyed students report that a university whose political atmosphere differed strongly from their own would make them hesitate to attend. We then combine the two data sources, using the conjoint estimates to correct the administrative estimates for endogeneity. Counterfactual simulations quantify the effect of a taste for political alignment: it leads students to attend programs 2.6 kilometers farther from home. The effects are heterogeneous by academic grade, socio-economic background, and political leaning: it pulls centrist students toward more distant, more expensive, and more selective programs, and left-leaning students in the opposite direction.

Political parties unite politicians of differing ideologies, yet incentives to follow the party line may limit how visibly these differences are expressed. We study whether within-party ideological divisions among U.S. House candidates are reflected in their language on Twitter, comparing personal and campaign accounts with official accounts. An ideological split explains how co-partisans speak better than most alternative partitions, especially on personal and campaign accounts (98th percentile for Democrats, 77th for Republicans). On official accounts, the relationship weakens for Democrats and disappears for Republicans, consistent with party pressure pulling institutional communication towards the party line.