Our October polls
Democrat-versus-Republican races only. Already voted is self-reported, not official returns. Leads are point estimates; a few are within the margin of error.
These findings combine answers from our separate state and district polls. They describe voters in those polls, not the whole country. Margins of error are at 95% confidence; smaller groups have wider margins. How we poll
Likely voters in the named October state and district polls; not a national sample or forecast. Pooled figures are respondent-proportional descriptive summaries of separate electorates. Each poll’s subgroup weights are normalized to its answered-question n; repeated people remain in record n but their weights are collapsed for Kish effective n. MOEs are approximate pointwise 95% half-widths in percentage points, without adjustment for exploratory multiple comparisons. They do not include all coverage, nonresponse, recall, calibration or selection uncertainty. Wald intervals can extend outside percentage bounds; an observed zero is not a population zero. Unknown and indifferent answers remain in denominators. Fewer than 50 distinct people is suppressed; 50–99 means interpret with caution.
