Each election year, the Cooperative Election Study (CES) produces summary results from its pre-election interviews with American adults. The 2024 pre-election results, including dynamic crosstabs, can be found here. Results for key swing states can be found here. The survey shows Kamala Harris with a 4-point lead over Donald Trump among likely voters (51% – 47%). About 3% of likely voters remain undecided.
The CES is the largest academic survey focused on American elections. Since 2006, the study has interviewed well over a half-million Americans to capture their views on contemporary policy debates, their engagement in political and social life, and their vote choices in federal and state elections. The study is a collaborative enterprise partially funded by the National Science Foundation and involving the participation of hundreds of scholars and students across dozens of academic institutions across the country. The survey is administered by YouGov.
During election years, the survey consists of interviews with more than 50,000 American adults before and after the election. The data from each study is made publicly available to scholars, analysts, and the public once the data is fully cleaned and processed several months after the election. These pre-election estimates are based on an initial data collection and cleaning, weighting, and the application of a likely voter model. However, the final dataset will include more rigorous data cleaning and weighting and a match of respondents to voter files to identify which individuals actually turned out to vote.
These pre-election estimates are based on preliminary data from the 2024 pre-election survey, fielded from Oct. 1st – Oct. 25th, 2024. The sample size for these estimates is 78,247 American adult respondents. Post-stratification weights were implemented to make the sample nationally representative of American adults by gender, age, region, education, race, the interaction of education and race, home ownership, and 2020 presidential vote. Adult population targets defined using the most recent 5-year American Community Survey. Probabilistic likely voter weights are applied to make estimates for likely voters using a model described here. These probabilistic weights imply a sample of 48,732 likely voters.
You can find trends from our surveys going back to 2008 here.