What this poll is
A rolling cross-sectional poll of the United States, following the race for the next President: which party's primary people plan to vote in, who they would vote for and who they would consider, and the views behind those choices. Fresh samples of American adults are surveyed online continuously, and the results update every hour as answers arrive, so opinion can be tracked as the candidates and the news move.
How the data is collected
Respondents come from established online research panels. Only people located in the United States are included. Each person answers a short set of topical questions followed by the political and demographic questions used for weighting. Every result on this site shows the dates and times of the responses behind it.
Quality checks
Unreliable responses are removed before any results are calculated. A response is removed if it:
- fails an attention check (an instruction hidden among a set of statements);
- is completed in under 40% of the typical (median) time;
- comes from outside the United States, is a duplicate, or never reaches the end of the survey.
Respondents who give the same answer to every statement, or whose typed answers are gibberish are flagged; their typed answers are not shown.
Weighting
Each question is weighted separately, on the people who were asked it, so that they match the adult population of the United States. The weighting (raking) adjusts for:
- age and gender together, region (Northeast, Midwest, South, West), race and Hispanic origin, education and household income;
- vote in the 2024 and 2020 presidential elections.
Smaller samples are weighted on fewer, broader variables: a simple set below 150 respondents and the full set from 150, with the two blended smoothly as the sample grows. Weights are capped so that no single respondent makes up more than about 3% of a result, which keeps the estimates stable.
Where the targets come from
- The American Community Survey – the Census Bureau's large, mandatory survey and the standard source for population figures – for age, gender, race and Hispanic origin, education and income; Census population estimates for region.
- Official election results for 2024 and 2020 presidential vote. Because online samples contain too few non-voters, non-voters are weighted part of the way towards their true share of American adults.
Primary voting intention, party identification and the most important issues are measured but not weighted on. Primary results are among everyone who says they would vote in that party's primary.
Reading the results
Each answer shows the weighted estimate and the uncertainty around it, based on the effective sample size after weighting. The shaded band is the 90% range, and the 20 lines are equally likely values that together span the 95% range – smaller samples spread out more. Results from fewer than 50 respondents are marked as small samples. Open “Breakdowns” under any question for results by age, gender, 2024 vote, ethnicity, region and other groups; groups with fewer than 30 respondents are greyed out.
Open-ended questions
Typed answers are grouped into themes, and each theme's share is weighted in the same way. The answers shown on the page pass an automatic filter for personal details and bad language before they appear.
Archive
When a question is retired it moves to the Archive, with results covering the whole time it was in the field, weighted in exactly the same way.