Controlling Elections
Should Moving to the Middle Win Candidates Votes? It Depends Where Voters Are
David Broockman & Joshua Kalla
Yale Working Paper, March 2026
Abstract:
Research finds that moderation only weakly improves candidate vote shares, which some argue indicates voters fail to vote on issues. We argue these findings can instead reflect two features of public opinion: party positions are popular on some issues, and public opinion is multidimensional. We show why, given these features, candidates' electoral returns from moving towards the other party's positions can be small even if all voters vote on issues. We illustrate this using a conjoint experiment (N = 6, 000) with presidential candidates and randomized policy positions. Movement towards the other party improves vote shares when party positions are unpopular, but produces null or negative effects when party positions command majority support. Candidates moving towards the other party win some voters but lose others who preferred their party's position -- producing small aggregate effects but consistent with widespread issue-based voting. Small measured effects of moderation may understate how many voters vote on issues.
The Effect of the Texas Migrant Busing Program on the 2024 U.S. Presidential Election
William Scarborough, Ronald Kwon & David Brady
Sociological Science, March 2026
Abstract:
From 2022 to 2024, Texas transported more than 100,000 migrants from the U.S.-Mexico border to six cities led by Democratic mayors, creating a unique migration shock far from the border. We use county-level data to estimate the program's effects on presidential elections. Comparing two elections prior to the program (2016-2020) with one after (2024), we find that the busing program increased Trump's vote share by more than three percentage points in treated counties. These effects are robust to alternative analyses. To explore mechanisms further, we analyze individual-level data from AP VoteCast. The increase in Trump's vote share in places receiving buses was driven by swing voters and elevated Republican turnout. Swing voters in busing destinations were moved to Trump by amplified concerns with crime, whereas Republican turnout was linked to heightened concerns over immigration. Our findings highlight the enduring power of minority threat and the growing role of subnational immigration policies.
Geography of Grievance: Industrial Hubs Magnify Political Discontent
Sung Eun Kim & Krzysztof Pelc
International Organization, Winter 2026, Pages 1-37
Abstract:
Why do some economic shocks have political consequences, upturning elections and ushering in radical candidates, while others are brushed off as structural change? We address this puzzle by looking to geographically concentrated industries, and how they relate to regional identity. While most often presented as a source of regional strength, we show that industrial hubs in the United States have accounted for more job losses than gains over the last twenty years. We then show how this matters through three original survey studies. Workers in geographically concentrated industries belong to denser, more deeply-rooted peer networks; these are associated with a stronger view that politicians are responsible for preventing layoffs. Those same individuals also perceive economic shocks of equal magnitude as more damaging to their region's standing, compared to the rest of the country. Perceptions of lost regional standing, in turn, are associated with greater demand for populist leadership traits. Finally, we show how these individual attitudes translate into aggregate political behavior. Employment losses in industrial hubs are tied to greater support for Republican candidates, while equivalent losses in non-hubs show no analogous effect. Our account presents a competing picture to the dominant narrative of industrial hubs as founts of innovation and productivity. When threatened by structural forces, such hubs can turn instead into founts of political resentment.
Politics transformed? Electoral competition under ranked choice voting
Peter Buisseret & Carlo Prato
American Journal of Political Science, forthcoming
Abstract:
We compare multicandidate elections under plurality rule versus ranked choice voting (RCV). We examine a widely held presumption that RCV more effectively incentivizes candidates to pursue broad campaigns that can appeal to all voters, rather than targeting a narrow segment of the electorate. That presumption is correct when preference transfers are competitive, that is, when multiple candidates have a reasonable chance of securing voters' second-choice support. However, when transfers are uncompetitive due to partisan, ethnic, or cultural alignments, that presumption is reversed: RCV can strengthen candidates' incentives to pursue targeted campaigns.
The Unequal Challenge of Learning from Under-Informative News
Andrew Trexler
British Journal of Political Science, February 2026
Abstract:
Political news consumption is highly uneven today: few people consume from news outlets directly, while many encounter news incidentally through social media and aggregators. Because outlets depend on direct consumers for revenue, they respond primarily to this core audience's preferences. Several contemporary styles of news coverage -- which emphasize partisan conflict, employ specialized jargon, engage in predictive analysis, and use clickbait language -- are attractive to core consumers, but may also make news less accessible for others. In a pre-registered survey experiment (n = 2,233), I show that, relative to a public interest style that prioritizes key information about policy and democratic norms, typical news styles weaken post-exposure recall of key news information -- that is, they are under-informative. Recall penalties are especially severe for those with lower baseline political engagement, yet still affect highly engaged consumers as well. This study shows that contemporary approaches to news coverage broadly under-serve the public by inhibiting political learning.
Why the Wait? Preprocessing, Mail Ballot Usage, and Vote Reporting Dynamics in the 2020 and 2024 Elections
Sina Shaikh & Charles Stewart
MIT Working Paper, January 2026
Abstract:
Public debates about 'found' ballots, partisan blue shifts, and election-night counting speed presume that election administration policies -- particularly preprocessing and mail-voting restrictions -- substantially affect how quickly and predictably results are reported. We analyze time-stamped election returns from 50 states and D.C. in 2020 and 2024, paired with state preprocessing rules and mail-ballot usage estimates. For each state-year, we measure votes reported and absolute deviations from final two-party vote shares at five benchmarks (1 hour to 1 week after polls close). State fixed-effects models reveal two patterns: mail-ballot usage is associated with slower early reporting but not with larger blue shifts or delayed convergence; preprocessing adoption shows more consistent associations with smaller deviations from final vote shares, though its effect on raw reporting speed is modest and sometimes statistically uncertain. A Pennsylvania counterfactual illustrates how small percentage-point changes can meaningfully affect the timing of outcomes in competitive states.
Encouraging crossover voting in the 2024 presidential primary
Hayley Cohen & Daniel Markovits
American Journal of Political Science, forthcoming
Abstract:
Will voters participate in the primary of a party they oppose to prevent the nomination of a candidate they fear? Partnering with a political action committee, we conduct a first-of-its-kind, large, preregistered field experiment (N = 83,902) in the lead-up to the 2024 Republican presidential primary in New Hampshire. A specialized get-out-the-vote intervention increases turnout in the Republican primary among undeclared voters who are modeled as likely to vote for Democratic candidates in the general election. Our treatment increased Republican primary turnout in this sample by 1.6 percentage points while reducing turnout in the Democratic Primary by .5 percentage points. Supplementing our experiment with surveys before and after the primary, we estimate that each vote cast by Democratic-leaning voters in the Republican primary had between a 78% and 95% probability of supporting the relatively moderate Republican primary candidate. We argue that voters are capable of sophisticated, risk-mitigating behavior in primaries.
The opportunities and limits of microtargeting cross-pressured voters
Philip Moniz, Kyle Endres & Costas Panagopoulos
Research & Politics, March 2026
Abstract:
Technological advancements have ushered in the promise of targeting potentially persuadable voters with personalized messages. One class of persuadable voters is those who are cross-pressured: they identify with one party but agree with policy positions of another. For political campaigns, targeting cross-pressured voters usually involves working with commercial vendors to infer voters' party identity and policy positions. The effectiveness of microtargeted messages thus depends on the accuracy of those predictions. We analyze the accuracy of a leading US commercial vendor's predicted policy positions using an original 2017 survey (N = 397) asking voters their actual positions on 20 issues. We find a strong negative relationship between cross-pressure and predictive accuracy, at both the issue- and individual-level. Issues on which voters are cross-pressured are predicted with less accuracy. And the percentage of correct inferences about a voter's positions decreases the more cross-pressured they are. While persuadable voters are a valuable target, reaching them is complicated by the difficulty in predicting their party-atypical stances.
Descriptive or Partisan Representation? Examining Trade-Offs for Asian Americans
John Cho, Mia Costa & Yusaku Horiuchi
British Journal of Political Science, February 2026
Abstract:
Do voters want representatives who share their race, ethnicity, or partisanship? We examine this question with a focus on Asian Americans who face trade-offs between descriptive (that is, Asian American or 'pan-ethnic') and partisan representation, as well as trade-offs involving 'co-ethnic' (for example, Korean for Korean) and 'cross-ethnic' (for example, Indian for Korean) descriptive representation. Across two experiments, we find that when Asian Americans are asked about collective representation in Congress, they prioritize more co-ethnic and pan-ethnic legislators over co-partisan legislators. However, in a competitive electoral setting, they often trade off race/ethnicity for partisanship. Asian Americans are only willing to cross party lines to vote for a co-ethnic candidate, but never for a cross- or pan-ethnic candidate. These findings shed light on the importance of considering heterogeneous preferences along ethnicities within the same racial 'in-group', such as Asian Americans, a heavily understudied and heterogeneous group in American politics.
Legibility and the Limits of Machine Politics
Amy Basu
Yale Working Paper, January 2026
Abstract:
The transition from machine politics to populism is often framed as a failure of democratic norms. Contra conventional accounts focusing on leader degradation or democratic backsliding, we propose a structural alternative based on organisational legibility. Our paper models the electoral portfolio allocation of a strategic party where the cost of monitoring contingent exchange is a convex function of the size of the informal periphery. As urban migration expands the "illegible" sector, the machine faces a liquidity crisis: the transaction costs of verifying votes exceed the campaign budget. The incumbent is compelled to abandon the retail politics of the broker for the wholesale politics of the coordination signal. We define this threshold as the machine's solvency horizon. The model demonstrates that the "hollowing out" of traditional parties is an endogenous response to the information asymmetry engendered by rapid urbanisation, challenging the view that populism is merely a preference shock.