2026 Theses Doctoral

Strategic Motivations and Conflict in American Democracy

Markovits, Daniel B.

This dissertation investigates how Americans perceive violations of democratic norms during a period when they are increasingly in question. I examine this question through three papers that focus on distinct but related beliefs and behaviors. In Chapter 1, I investigate expectations of retaliation for violations of democratic norms.

To what extent can fear of partisan retaliation deter support for anti-democratic behavior? In contemporary American politics, many voters worry about the opposing party’s commitment to democracy. Such concerns could, in theory, promote compromise if voters believe opponents will break democratic rules only when provoked. I investigate these beliefs among the American public, distinguishing them from other drivers of democratic support such as meta-perceptions of public opinion. In two prediction experiments (N = 7,000, 29,000 observations), I find that partisans view the opposing party as only modestly more likely to undermine democracy if provoked than if unprovoked.

These expectations fall well below both theoretical maximums, with that gap robust to many different approaches to framing the results. Further, unprompted references to retaliation are rare across corpura of open-ended responses from both survey and social media data. In follow-up experiments (N = 5,500), I show that randomized warnings about the likelihood of retaliation reduce support for violating democratic rules. These results demonstrate that while strategic reasoning is limited at baseline, simply priming the risk of retaliation can reduce support for anti-democratic behavior. In Chapter 2, I explore beliefs about the effectiveness of democratic backsliding.

What do Americans believe are the consequences of violating democratic rules? Recent scholarship shows that Americans broadly oppose anti-democratic behavior, yet are often unwilling to sacrifice partisan goals to defend democratic norms. Despite a growing literature on public attitudes toward democracy, we know little about whether citizens view such behavior as effective in shaping electoral outcomes or enabling otherwise unattainable policy goals. I address this question using an incentivized prediction experiment conducted in the lead-up to the 2024 presidential election, along with a follow-up survey experiment. I find that partisans believe anti-democratic behavior can influence electoral outcomes, with co-partisans of the violating party estimating much smaller effects and little distinction between the effectiveness of different types of violations. In contrast, partisans of both parties estimate null effects of democratic violations on policy outcomes.

My results suggest that partisans hold realistic and modest assessments of democratic violations but tend to exaggerate the effects of the opposing party's efforts at electoral manipulation, a dynamic that could fuel escalating conflict over democracy in the electoral arena. In Chapter 3, I investigate a type of behavior with new salience in an era of rising polarization and democratic threat: crossover primary voting.

American primary elections allow for partisans of the opposing party to participate. Do they? Using data from multiple national surveys spanning 10 elections, we show that between 3-5% of existing primary voters identify as members of the opposing party, and so engage in ``crossover voting.'' These voters meaningfully moderate the primary electorate. Then, we use original surveys (N = 2,422), three field experiments (N = 237,849), and a modified conjoint experiment (N = 2,055) to explore why this crossover voting remains rare.

We highlight the role of (1) psychological costs, including normative beliefs about the appropriateness of crossover voting and negative affect towards the opposing party, (2) electoral context, including whether candidates in the opposing primary are meaningfully differentiated and whether the general election is expected to be competitive, and (3) the lack of campaign contact, which could provide eligibility and candidate information as well as social sanction. By experimentally varying all three dimensions, we observe substantial variation in behavioral and attitudinal measures of crossover voting. Our results suggest that combinations of behavioral interventions and changes to the pool of primary candidates could dramatically re-shape primary electorates in favor of ideological moderation and that crossover voting near-universally favors moderate candidates.