June 10, 2025, 3:47 p.m.
Issue polls sometimes get a bad rap. Indeed, it’s challenging to measure what people think about complex policy areas that they may rarely think about. Issue polls are really delving into how politics intersects with people’s daily lives.
They’re also what I love most about being a pollster. Election polls are fine, but asking who someone will vote for is just about the easiest and most boring thing a pollster will ever do. (Whether they will vote is trickier.) I got into polling because I want to know what in politics makes people tick—or doesn’t—beyond their vote choice.
In most non-election years, issue polls often don’t get a ton of press—it’s not a game-ified horse race, so fewer people care. In 2025, though, we are seeing tons of coverage of issue polls. Everyone wants to know whether President Trump’s policies are popular.
Republicans and the Trump administration want to claim the majority supports what they’re doing. Democrats want to be able to say the majority opposes Trump. Right now, you can find polls that support both arguments. To take just one example: The broad strokes of Trump’s immigration agenda are popular, while many of the specific actions he is taking are not.
Which matters more depends on your purpose in analyzing the polls. If you’re thinking that the next election will hinge on swing voters who aren’t paying close attention, you might care more about the broad strokes. If you’re making an argument about the specific policies, that data is more useful.
But do we even know if we’re measuring those opinions accurately?
We don’t. There won’t ever be an election on most issues where we get a “truth” which we can compare to poll results. And there’s reason to be cautious in exploring opinion on Trump’s agenda: Pollsters systematically underestimated Trump's support in 2016 and 2020, and to a lesser extent in 2024. Does that mean we’re underestimating support for his policies as well?
To answer that, we have to look at how the key sources of election-poll problems—who answers polls and who actually votes—affect issue polling. The latter is simple to dismiss; the vast majority of issue polls include all registered voters, or even all American adults. That removes the need to jump through fire-rimmed hoops of figuring out who will turn out to vote.
The question of who answers polls, however, is just as fraught for issue polling as it is for election polling. Systematic nonresponse doesn’t just affect election toss-ups. We will still naturally underrepresent Trump supporters unless pollsters take steps to adjust their data. Fortunately, most pollsters now do take those steps. Moreover, it’s simpler to make sure we have enough Trump supporters in the data when we’ve just had an election that told us how many to expect.
Support for policies is also highly dependent on how the question is worded. I encourage anyone looking at an issue poll to go find the original question wording, because even minor shifts can change how people respond, particularly when people might not know much about the details of a policy. Pollsters can intentionally frame the question a certain way if they want to nudge respondents to answer in a certain way.
Answer options matter, too—specifically whether we give respondents a “don’t know” option. In a striking example, The New York Times reported that some polls showed 58-60 percent of respondents disapproved of the Department of Government Efficiency, but that dropped to 44-48 percent in polls that offered a “don’t know” response option.
One more consideration: People might not want to admit they support some of Trump’s policies. There’s not a ton of concrete evidence that people shy away from admitting they support Trump or his policies, but it can happen. Some studies do show that people are more likely to support some types of actions if they don’t have to explicitly say it.
Fortunately, if polls are underestimating support for Trump policies, it is likely only by a few percentage points, at least in high-quality polls that are trying to write fair questions, give fair answer options, and ensure they are getting the right mix of respondents. As always, though, beware of crappy polls that don’t tell you how they got their numbers or how they worded their questions.
Contributing editor Natalie Jackson is a vice president at GQR Research.
