Forecasters predict Democrats to take House but race for Senate hard to call
In a forecasting special issue of PS: Political Science & Politics for the US midterm elections, all forecasters predict that the Democrats will capture the House of Representatives, but are split on who will emerge victorious in the Senate.
Published by Cambridge University Press on behalf of the American Political Science Association (APSA), the special issue includes 17 studies showcasing a range of forecasting approaches and methodologies, from traditional political economy models to controversial betting markets and AI ‘synthetic personas’.
The US Congressional Midterm elections take place on 3 November 2026. They will see all 435 seats in the House of Representatives and 35 Senate seats contested, with control of Congress potentially decided by a handful of highly competitive races.
More broadly, the collection explores how political dynamics are reshaping both elections and the tools used to understand them, and asks whether traditional models can still make accurate predictions in a time of heightened polarisation and political change.
Dr John Kenny, co-Guest Editor of the special issue from the University of Mannheim and University of Southampton, said: “Midterm elections have historically seen the president’s party lose support, driven by economic concerns and falling approval ratings.
“But the political landscape in 2026 is somewhat atypical. Increased polarisation and widespread redistricting have introduced new uncertainties, creating challenges for predicting whether this election will follow established patterns or whether it will mark a significant shift in American politics.”
Predictions for 2026
Many of the studies in the special issue have made predictions on the outcomes of the midterm elections taking place on 3 November 2026.
The verdict was unanimous on a Democrat victory in the House race, with 12 out of 12 predictions in their favour.
Predictions for the Senate race were less clear. Some forecasts are calling it for the Republicans and others are predicting a Democrat victory. All are expecting a close and contentious electoral victory.
Dr Debra Leiter, co-Guest Editor from the University of Missouri-Kansas City, said: “With such a narrow Republican majority, Democrats require only modest gains to retake the House, but the Senate will be harder to flip due to the number of Republican-held seats considered secure. Record breaking spending, open-seat contests created by a wave of retirements, and unexpected candidate choices are expected to increase competitiveness across key battleground states.
“Coupled with the unique circumstances of the 2026 election, it’s a tough one to call. Will we see a return to the fundamentals of American election forecasting, or will it prove a challenge to the widespread beliefs about American electoral behaviour?”
Highlighted papers
Roberto Cerina. ‘Polling Without People: Can Silicon Samples of CASM-Inspired LLM Personae Mimic Human Respondents in Pre-Election Polling?’
Contact: Dr Robert Cerina, University of Amsterdam
r.cerina@uva.nl
Roberto Cerina explores whether AI-generated ‘synthetic personas’ can replicate human responses in pre-election polling. Cerina created detailed large language model personas based on factors such as age, gender, education and political identity.
The model’s key innovation is its ability to simulate media consumption. Using up-to-date news feeds, personas were exposed to current events and asked how those developments influenced their political preferences. This helps to overcome a limitation of conventional AI models, whose knowledge is restricted by training cut-off dates.
The model forecasts a Democratic victory in the House, with 58.4% of the national vote and around 292 seats. Though the forecast may overestimate Democratic support, the approach offers a transparent way to understand why voters change their preferences. The paper argues AI polling could be a complement to traditional surveys, offering more frequent, granular insights into public opinion.
Thomas Gruca and Thomas Rietz. ‘Forecasting 2026 Congressional Control: A Comparison of IEM and Kalshi Prediction Markets’
Contact: Professor Thomas Rietz, University of Iowa
thomas-rietz@uiowa.edu
Thomas Gruca and Thomas Rietz examine the Iowa Electronic Market (IEM), a classic futures election market, alongside Kalshi, one of the fastest growing online betting markets. They demonstrate the stability of IEM forecasts, in contrast to the greater instability of the Kalshi markets.
In separate markets, the IEM predicts that the most likely outcomes are a Democratic House victory and a Republican Senate majority. Kalshi markets predict the most likely outcomes are a Democratic House and Senate majority, but less consistently. In combined control markets, both indicate the most likely joint outcome is a Democratic sweep. While both produce similar predictions, the IEM markets have an established history of effective and reliable forecasts that may be less prone to certain kinds of market manipulation.
The most recent IEM prediction gave a 60 per cent chance that Democrats would take the House.
Kalshi predicted similar outcomes, but the chance of the Democrats taking the House was significantly higher at close to 90 per cent.
Both markets predict a Democratic Senate.
Peter Enns, Leigh Farah, Claudia Miner, Thomas Gareau-Paquette. ‘A District-Level Forecast of the 2026 U.S. House of Representatives Midterm Elections’
Contact: Professor Peter Enns, Cornell University
pe52@cornell.edu
Peter Enns et al. recognise that observers of the US midterm elections are likely focused on who will win the majority, yet 435 separate district contests will determine political authority in the House. Their model takes key election factors in each of these districts, including this year’s unique mid-decade redistricting, candidate success measures such as fundraising, and both national level and state level vote intention, to produce a forecast model that mixes national and localised information together in a new and effective way.
Based on 10,000 simulations using data from 100 days before the election, the forecast gives Democrats a 90 per cent chance of taking the House, with an expected total of 235 seats.
Charles Tien and Michael Lewis-Beck. ‘Political economy forecasts: the 2026 midterms’
Contact: Professor Charles Tien, Hunter College, CUNY
ctien@hunter.cuny.edu
Professor Michael Lewis-Beck, University of Iowa
michael-lewis-beck@uiowa.edu
Charles Tien and Mike Lewis-Beck unveil a new forecast of the 2026 elections based on the long-running Political Economy Model. The model draws on three key predictors: economic performance, presidential approval ratings and the historical impact of midterm elections. The paper argues that voters reward governing parties when conditions are favourable, and punish them when they are not.
With President Donald Trump’s approval rating at around 35%, the model predicts Republicans will lose 33 House seats and lose their narrow majority. It also predicts the Democrats will pick up enough seats to retake the Senate. The forecast reflects the powerful historical tendency for presidents’ parties to suffer losses at midterms.
Stephen Quinlan and Michael Lewis-Beck: ‘Demslide or Republican resilience? A political history forecast of the 2026 US congressional elections’
Contact: Stephen Quinlan
stephenpfquinlanwork@gmail.com
Professor Michael Lewis-Beck, University of Iowa
michael-lewis-beck@uiowa.edu
Drawing on decades of US political history, Stephen Quinlan and Mike Lewis-Beck have developed a forecasting model that examines the structural and historical conditions that have consistently influenced election outcomes over time. When one party controls the presidency, Senate and House simultaneously, voters have tended to impose an electoral penalty on the governing party.
With Republicans currently controlling all three branches, the model suggests conditions are favourable for Democratic gains. However, increasing political polarisation and a declining number of competitive House districts may limit gains. The researchers' central forecast projects Democrats winning 231 House seats and 50 Senate seats, giving them a 78% chance of recapturing the House and a 66% chance of winning control of the Senate.
Jack Rametta and Christopher Hare: ‘How Model Choice Obscures Forecast Uncertainty: The Rashomon Effect and the 2026 Senate’
Contact: Dr Christopher Hare, UC Davis
cdhare@ucdavis.edu
Jack Rametta and Christopher Hare have developed a forecasting model to predict US Senate election outcomes that reflects the growing nationalisation of congressional elections, where local candidate appeal and the incumbency advantage are increasingly outweighed by broader political forces such as presidential approval, partisanship and economic conditions.
The researchers argue that elections that were once shaped more heavily by state-level dynamics are now closely tied to voters’ perceptions of national political performance. As a result, even well-regarded incumbents can struggle to withstand unfavourable national trends.
As of late September, the forecast predicts Democrats will take 50 seats in the Senate, but given the high uncertainty, the forecast ranges from 45 to 56 seats. Democrats are slightly more likely than not to secure a Senate majority, though the race remains highly competitive.
Olivia Quinn, Amanda Brush and Eric R. A. N. Smith: ‘Updating the Fundamentals: Forecasting the 2026 Midterm Election Post Gallup’
Contact: Dr Olivia Quinn, UC Santa Barbara
oquinn@ucsb.edu
The recent discontinuation of Gallup’s Presidential Approval polls, an 88-year tradition, poses a problem for many election forecasting models. Olivia Quinn, Amanda Brush and Eric R. A. N Smith show that these models can continue without it. They develop a proxy measure from RealClearPolitics’ poll average and use it to predict what will happen on election night.
As of the most recent quarter, their Congressional seat change forecast predicts sizable Democratic gains in both chambers, with Republicans predicted to lose nine seats in the Senate and 42 seats in the House.
Notes to Editors:
For general enquiries about the issue, please contact:
Dr Matt Midgley, Cambridge University Press & Assessment
+44 (0)1223 342979
To speak to the guest editors, please contact:
Dr Debra Leiter, University of Missouri-Kansas City
leiterd@umkc.edu
Dr John Kenny, University of Mannheim/University of Southampton
J.P.Kenny@soton.ac.uk
Publication:
The Special Issue Forecasting the 2026 US Midterm Elections of PS: Political Science & Politics is published on behalf of the American Political Science Association by Cambridge University Press.
All papers in the special issue will be available here when published: https://www.cambridge.org/core/journals/ps-political-science-and-politics/accepted-manuscripts
Studies listed below and highlighted in this press release are strictly embargoed until Thursday 15 October 2026 at 00:01 UK time / 14 October at 19:01 US Eastern Time.
- Roberto Cerina. ‘Polling Without People: Can Silicon Samples of CASM-Inspired LLM Personae Mimic Human Respondents in Pre-Election Polling?’.
https://DOI.org/10.1017/S1049096526103084 - Thomas Gruca and Thomas Rietz. ‘Forecasting 2026 Congressional Control: A Comparison of IEM and Kalshi Prediction Markets’. https://DOI.org/10.1017/S1049096526103023
- Peter Enns, Leigh Farah, Claudia Miner, Thomas Gareau-Paquette. ‘A District-Level Forecast of the 2026 U.S. House of Representatives Midterm Elections’. https://doi.org/10.1017/S1049096526103072
- Charles Tien and Michael Lewis-Beck. ‘Political economy forecasts: the 2026 midterms’. https://DOI.org/10.1017/S104909652610290X
- Stephen Quinlan and Michael Lewis-Beck. ‘Demslide or Republican resilience? A political history forecast of the 2026 US congressional elections’. https://DOI.org/10.1017/S104909652610300X
- Jack Rametta and Christopher Hare: ‘How Model Choice Obscures Forecast Uncertainty: The Rashomon Effect and the 2026 Senate’. https://doi.org/10.1017/S1049096526103059
- Olivia Quinn, Amanda Brush and Eric R. A. N. Smith: ‘Updating the Fundamentals: Forecasting the 2026 Midterm Election Post Gallup’. https://DOI.org/10.1017/S1049096526103047
About Cambridge University Press
Cambridge University Press is the academic and Bibles publisher of Cambridge University Press & Assessment. It publishes books and journals, serving customers in higher education through research, education products and services. It is part of the University of Cambridge, delivering trusted research and learning materials that spread knowledge, spark curiosity and aid understanding of the world we live in.
cambridge.org/universitypress
About American Political Science Association (APSA)
Founded in 1903, the American Political Science Association is the leading professional organization for the study of political science and serves more than 11,000 members in more than 80 countries. With a range of programs and services for individuals, departments, and institutions, APSA brings together political scientists from all fields of inquiry, regions, and occupational endeavors within and outside academe to deepen our understanding of politics, democracy, and citizenship throughout the world.
No comments:
Post a Comment