One of the things about politics that people love to hate is polling. To a certain extent, this antipathy is entirely understandable. As critics often say, political journalists sometimes overreport polling at the expense of candidate and party-issue positions, policy debates, or elected-official conduct. Polling resources tend to be concentrated on horse-race measurements rather than gathering data on citizen preferences that inform elected officials and the public. Unprincipled politicians are tempted to promise whatever makes their polling numbers rise rather than what they actually think — or even know — is right. And polling contributes to the reduction of political strategy to micro-appeals to narrow swing-voter groups rather than the development of broad sustainable coalitions.

All of these complaints can be addressed, mostly by avoiding overreliance on polls and keeping some perspective about what they can or cannot tell you. But there’s one complaint that’s less compelling, and we are hearing it a lot in 2026: The polls aren’t accurate enough.

This objection emanates mostly from people who only want to know who’s going to win and lose elections, so any data that gets that “wrong” is useless. In very close elections, of course, polls can be quite accurate and still fail to “predict” who is going to win (not that pollsters typically make predictions; that’s the job of prediction markets!). Indeed, polls are invariably accompanied by margins of error for both top-line and cross-tab findings that make the variability of the numbers very clear. But that gets pushed aside when gloating winners or angry losers choose to trash the polls for not being precisely accurate. In the former category is Donald Trump, who has his own sinister reasons for undermining the credibility of polls. He complained on Truth Social earlier this year:

Fake and Fraudulent Polling........

© Daily Intelligencer