Our Methodology

The Honest Poll conducts surveys of Likely Voters (LV), Registered Voters (RV), and General Adults depending on the nature of the question being measured. For electoral polling, we default to Likely Voter samples using a proprietary screening methodology refined over years of national polling.

Sample sizes are determined by the desired margin of error and population being measured. Standard national polls use samples of 1,000–1,500 respondents, yielding a margin of error of approximately ±3 percentage points at the 95% confidence level. State-level polls typically use 700–1,000 respondents, sized to the target population and required precision. Congressional Districts, and similar, are examined on a case-by-case basis.

All samples are drawn to be representative of the target population by age, gender, race/ethnicity, geography, and party identification. Demographic weighting targets are sourced from the U.S. Census Bureau’s Current Population Survey (CPS) and American Community Survey (ACS).

Surveys are administered using the methodology most appropriate for the engagement — including online panels, SMS-to-web, telephone (IVR and live), voter-file matched samples, and opt-in panel recruitment depending on the target population and research objectives. Mode is always disclosed in published methodology. We do not rely exclusively on opt-in online panels for political polling, as these can introduce selection bias.

Fieldwork windows are published with every poll release. We do not compress fieldwork into unrealistically short windows that could bias results toward respondents who happen to be available during specific periods.

Surveys are administered in English as standard. Spanish-language versions are available upon request for bilingual or Hispanic-oversampled engagements. Language version used is noted in published methodology.

Raw samples are weighted to match known population parameters. Weighting targets include age, gender, race/ethnicity, education, and geography, sourced from established population research appropriate to the engagement. Party identification weighting uses recent electoral results as a calibration target — a more predictive benchmark than self-reported registration, which tends to overstate Democratic affiliation in general population samples.

All published polls include the weighted sample composition. For public polling, only final weighted results are published. Private commissioned polling deliverables may include additional internal breakdowns as agreed with the client.

Every question is written in good faith to measure opinion accurately. We do not use deliberately misleading premises, deceptive framing, or question ordering designed to manipulate respondents. Full question wording and question order are always published verbatim — readers can judge the work for themselves. All non-erroneous results from public polls are published in full. We do not suppress, selectively release, or sit on findings because they are inconvenient.

Private commissioned engagements may include persuasion-testing or message-framing questions at the client’s request. Clients provide topics and objectives; we write the questions. Complete question wording and order are always disclosed to the client. Private results are delivered in full to the client and are never selectively published to mislead the public.

Question wording and question order are published verbatim with every public poll release. Private commissioned deliverables include full question text and order provided to the client.

Every public poll release includes: sample size, fieldwork dates, population surveyed (LV/RV/Adults), geographic scope, margin of error at the 95% confidence level, mode of data collection, question wording, weighting targets, and full methodology statement. Margins of error reflect sampling error only; non-sampling error sources such as question wording effects and coverage bias are addressed separately in the methodology. No exceptions.

Corrections are published prominently and immediately. Being wrong is not shameful. Hiding being wrong is. We apply this standard to our own work and to everyone else’s.