Do Primary Elections Really Fuel Congressional Polarization?: Professor Anthony Fowler Investigates

As voters across the United States head into the 2026 primary season, a familiar claim is once again taking center stage in political debate: that primary elections push elected officials toward ideological extremes and fuel polarization. A new study by Anthony Fowler, the Sydney A. Stein Jr. Professor at the Harris School of Public Policy, and Shu Fu of Shanghai Jiao Tong University offers the most direct and comprehensive test to date of that assumption—and arrives at a more nuanced conclusion.

“Members of the U.S. Congress do not closely represent the preferences of their constituents, with Republicans positioned consistently to the right and Democrats to the left of the median voter in their respective constituencies,” the authors write in the opening line of the paper. “Paradoxically, voters prefer more ideologically moderate candidates, so both parties could win more seats if they were willing to field more moderate candidates. Why, therefore, are our elected officials so extreme relative to their constituents, and why is there so much polarization in Congress?”

In the paper, “Do Primary Elections Exacerbate Congressional Polarization?,” which is forthcoming from the Journal of Politics, Fowler and Fu ask a simple but powerful question: do members of Congress vote differently when they are worried about winning their party’s primary election? The fear of being “primaried” is often cited by pundits and practitioners as a central driver of polarization, discouraging compromise and rewarding ideological purity. If that fear meaningfully shapes legislative behavior, it should be visible in how members vote before and after their primary elections.

To test this idea, the authors exploit a distinctive feature of the U.S. electoral system: primary elections occur on different dates across states. Taking this into account, Fowler and Fu analyze roughly eight million roll-call votes cast in the U.S. House and Senate between 1995 and 2022. Their research design compares how the same member votes before versus after their state’s primary election date, while carefully accounting for differences across legislators, parties, and individual bills.

The results show that primary elections do exert a measurable influence on congressional voting—but the size of that influence is modest. Members of Congress are slightly more likely to cast ideologically extreme votes before securing their party’s nomination. After the primary has passed, they vote somewhat more moderately. Taken together, however, these shifts explain only about one percent of the overall ideological distance between Democrats and Republicans in Congress.

The effects are not uniform. Fowler and Fu find that primary-related polarization is larger in the Senate than in the House, larger for those less likely to seek reelection, and more pronounced among relatively moderate lawmakers. The effects are weaker on party-priority legislation and in states that use nonpartisan primary systems, such as top-two or top-four primaries. These patterns align closely with theoretical expectations: primary pressure matters most where it is most salient—and least where electoral rules dilute partisan incentives.

Notably, the findings challenge the idea that primary elections are a dominant driver of polarization. Even among members unlikely to seek reelection again—those who should be least concerned about future primaries—the estimated effects remain small. Primary elections may nudge legislators toward ideological extremes at the margins, but they do not resolve the larger puzzle of why Congress remains so polarized despite the presence of broadly moderate voters.

The study also carries implications for ongoing debates about primary reform. While Fowler and Fu find suggestive evidence that nonpartisan primaries reduce polarizing incentives, they caution against interpreting their results as an argument against primaries altogether. Primary elections serve important accountability and representational functions, particularly in partisan districts. Their findings instead point to a more measured conclusion: how candidates are nominated can shape incentives and behavior, but primary elections alone are not the central cause of congressional polarization.

“Primaries matter, and do drive candidates to be more partisan, but most of the partisanship we see comes from elsewhere,” Fowler said.

These insights are especially timely as the 2026 primary calendar gets underway, including Illinois’s upcoming March 17 primary. As voters, candidates, and observers assess the health of American democracy, this research underscores the value of grounding political debates in careful empirical evidence rather than conventional wisdom.

Interested in learning more?

For those interested in learning more about Anthony Fowler’s research on political moderation, polarization, and electoral incentives, Fowler will be discussing his work at "Harris Up Front with Anthony Fowler" on March 12 at Maggiano’s in Chicago. The event offers an opportunity to explore what the evidence really says about polarization—and what it means for the future of American politics. Click here to register.