Voters should examine who conducted an election poll, how it reached people and whether its methods are transparent before treating its numbers as a prediction, Pew Research Center President Michael Dimock argues in an analysis of the 2026 midterms.
The analysis, republished by Pew on Oct. 29 after first appearing in The Hill on Oct. 14, describes rising costs, uncertain turnout and new AI-related risks. It is an assessment of polling, rather than a newly released voter survey.
Dimock's central distinction is between measuring what people think and predicting who will vote. Polls can estimate preferences, but turnout is especially difficult to anticipate in midterms, primaries and lower-information local contests.
Costs, respondents and turnout
High-quality polling has become more expensive, while cheaper opt-in approaches have brought more entrants into the field. Some have strong records, Dimock writes, but coverage and quality vary, particularly in local races.
He also points to paid respondents who misrepresent themselves and AI bots that can enter online surveys. Such problems can distort results, including estimates for younger and Hispanic respondents.
Changing voting rules and increasingly targeted campaigns add difficulty to predicting turnout. Pollsters must distinguish the electorate that could participate from the people likely to cast ballots.
New uncertainties for the midterms
Dimock identifies synthetic polls that do not interview humans, unreliable substitutes for real surveys and outright fabricated polls as additional concerns.
He also cites disputes over mail voting, voter intimidation, redistricting and a field with fewer incumbents, along with issues such as AI and data centers that may affect local contests without fitting neatly into national partisan divisions.
Those uncertainties make it risky to build expectations around one poll or an unusual result.
Dimock recommends looking at a pollster's track record, financial or political interests, question wording, sampling and disclosure of methods. Transparency helps readers evaluate a poll, but is not itself a guarantee of accuracy.
His conclusion is not that voters should discard polling. Carefully conducted surveys remain an important way to hear from the public, he argues, provided readers understand their methods and limits.
FingerLakes1.com is the region’s leading all-digital news publication. The company was founded in 1998 and has been keeping residents informed for more than two decades. Have a lead? Send it to news@fingerlakes1.com.
