Dem
52
Ties
3
GOP
45
Many people don't trust the polls. After all, polls predicted that Hillary Clinton would win in 2016. And we are not going to even start on Ann Selzer (who nailed it exactly for 12 years, got one poll wrong, and became a laughingstock as a result).
There is a core problem here and it is a psychology problem, not a math problem. American Airlines operates over 2 million safe flights a year, but the only one anyone remembers is the one in Jan. 2025 that collided with a helicopter in D.C. Same with polls. Most of them are right on target, assuming you understand what they are saying. If a poll says Smith is at 47% and Jones is at 44% the pollster is NOT saying Smith will win. The pollster is saying there is a 95% chance that Smith will end in the range 43% to 51% and there is a 95% chance Jones will end in the range 40% to 48% (assuming a 4% Margin of Error). If Jones wins 48% to 43%, the pollster will correctly say: "I nailed it" but many people don't understand that.
Since polling is rather "in the news" right now, as it were, Politico talked to four top pollsters to basically ask: "Why should anyone believe you?" Good question.
The pollsters polled were Seb Wride (PF Polling), Christine Matthews (Bellwether Research), Charles Franklin (Marquette Law School Poll), and Charlotte Swasey (Searchlight Institute). Matthews said that there are a lot of amateur (or worse) pollsters out there and they give the field a bad reputation (see above). In the past two midterms, the established pollsters actually did pretty well. She also said that some of the sources of problems in the past, especially non-response by Republicans, are being compensated for now. She also said all the talk of a realignment is premature. One election can't show that there is a fundamental realignment, as things don't change so fast.
Franklin said that methodology has changed a lot in the past 10 years. For one thing, they are using a much finer granularity for dividing the country into regions and matching results to known characteristics of each region. He also said that the polling is consistent in a lot of ways. People were unhappy with inflation under Joe Biden so they voted for Republicans in 2024. They are still unhappy with inflation and the polls show that now the Republicans are taking the brunt of it.
Swasey noted that the blocs are not fixed and the pollsters deal with that. She said that the Latinos who vote in presidential elections are not the same Latinos who vote in midterms and she can correct for that. She looks at trends to see where things are heading.
Wride said that when people tie specific things in their lives to specific actions by a politician or party, it affects how they vote. Right now, the increase in gas prices can be directly tied to the war in Iran. That is going to hurt Republicans.
The moderator asked if there is any demographic to watch closely this cycle. Wride said asking about MAGA identity is a good one. If that is slipping, it is a danger signal for the Republicans. Matthews said to look at the fraction of Trump voters who now disapprove of him. Franklin agreed with this and looks at Democrats who support the Republicans and vice versa as an important indicator. Swasey likes to look at how individual candidates are doing compared to the generic House poll. Greatly overperforming or underperforming the generic candidate is a clear signal about that race.
The moderator also asked whether prediction markets could replace polling. Wride said they are secondary. The bettors are just going by the polling. The prediction markets don't add anything new. Franklin noted that in the Wisconsin Democratic primary, the early frontrunner (who lost) was at 90% in the prediction markets until the actual election results started to come in. He doesn't think they add anything either. Swasey said the prediction markets are consumers of polls more than anything else.
The moderator wanted to know about the future of polling. Franklin said: "Oh, woe is us, because we have critics." He noted that journalists also have critics. Swasey said polling is the worst system—except for all the others that have been tried. Wride was optimistic about the future of polling. Despite a few high-profile misses, most polls have been pretty good. (V)