Every election cycle we invite readers to send in questions about how The New York Times is covering the campaign. This year, nearly 200 of you wrote to us, and polling was easily the No. 1 topic.
With polls showing Democrats favored to make a strong showing on Election Day, even in many reliably red districts and states, many of you wanted to know: Can the poll numbers be believed? Is support for President Trump and his party being underestimated, as it sometimes has been in the past? How do we ensure our methodology is sound and unbiased? There were also questions about where our political reporters are deployed and how they do their jobs, along with other topics.
We sent a sampling of some of the most common questions to editors in our Election Analytics department, which handles polls and election results; our Politics and National desks; and our Standards team. Here are their answers.
Changes to polling
How has the polling methodology changed in the last 10 years to avoid something like the 2016 presidential polling debacle? — Jack McCumber, Seattle
Ruth Igielnik, a Times editor who conducts polls and analyzes the results, answers:
After the 2016 election, industry pollsters did an extensive self-review and produced a wonderful but dense report. It found that one of the major challenges that pollsters faced was that Americans had not previously voted differently by education. But in 2016, many voters with college degrees voted for Hillary Clinton and many voters without a college degree supported Donald Trump, and polls at the time were not set up to correct for that difference. After 2016, many pollsters added information about voters’ education level to their statistical models to account for those shifts.
The other big change after 2016 was how pollsters talk about uncertainty. Polls come with many sources of error, which was not always made abundantly clear. Polls can tell us a lot, but they are not as precise as people would hope! (On that front, it’s worth noting that polling error was actually worse in 2020 than in 2016, but it got less attention because it erred in the “right” direction, with polls overestimating Joe Biden’s margin of victory.)
Pollsters also now contact voters via phone, text, email and snail mail. That is not to say that some of the tried-and-true methods for contacting voters do not still work. At the Times/Siena poll, we still primarily call voters.
But we are also always trying new things. For example, we now consider people who only complete part of our survey to have taken the survey, because we know that the types of people who complete a long survey are different politically from those who do not stay on the phone for as long.
Why is it that polls have been so wrong in the last few elections? Polls for states seem to be more inaccurate, sometimes even with pollsters who are respected in their fields. — Mary Ellen Giesey, McMurray, Pa.
Caroline Soler, a researcher focused on collecting and analyzing polling and election data, answers:
Polls can be wrong for a lot of reasons, and some elections are much harder to poll than others. Primary elections are especially difficult because turnout is unpredictable, and the races themselves are volatile. That helps explain why some polls in this year’s primaries missed by so much — a pattern similar to numerous recent elections.
That is somewhat different from the polling errors that received so much attention after the last three presidential elections, when polls routinely underestimated President Trump. Several theories have tried to explain why; one prominent explanation was that pollsters failed to reach disengaged, Trump-leaning voters. In response, many pollsters changed how they conducted their surveys by weighting their samples to match voters’ education levels and past voting behavior. These changes appear to have helped: 2024 had the lowest average presidential general election polling error in 20 years. (Yes, polls still underestimated Mr. Trump, but by less.)
Still, voters’ memories appear to be long and trust in polling has been slow to recover. A small number of bad actors touting fabricated “polls” as if they were real has only complicated matters further.
Accuracy of Times polls
How well do you think we can trust your poll results? — Gillette Hall, Santa Barbara, Calif.
Ruth Igielnik answers:
We can absolutely trust polls to tell us how Americans are feeling in this current political climate. In a recent poll we conducted in Ohio, voters were twice as likely to say that President Trump’s policies hurt their state rather than helped it, with 48 percent of respondents saying they had hurt Ohio and only 24 percent saying they had helped. (27 percent said his policies hadn’t made much of a difference.) Polls come with a margin of error — in this case it was 4.5 percentage points. That tells us the numbers could be about five percentage points higher or lower and still be accurate. For a finding like this, a result that was five percentage points different in either direction would not change the overall picture of how voters are feeling.
However, many of our elections are exceedingly close. In that same Ohio poll, we found that the Democrat Sherrod Brown had a narrow three-percentage-point edge over the Republican incumbent, Jon Husted, in their Senate race. That is within the margin of error. So effectively what that poll tells us is that the race is very close. Polls are imperfect tools to measure sentiment with precision, but they can give us a general sense of where things stand in a race and how voters are feeling.
Can you ensure an even spread of conservative, liberal and moderate voices in your polling? — Sarah Edwards, Gallatin, Tenn.
William P. Davis, the director of election analytics, answers:
Our polls start with a list of every registered voter in the country. We can use that list to make sure we are calling the right number of Democrats, Republicans and independents. After we get all the responses back, we use a statistical technique called weighting to make sure we have the right balance not just by political party but also by race, education, age and many other criteria.
Unfortunately, though, the party affiliation data in the voter file cannot tell us whether somebody is conservative, liberal or moderate. And all polls have a well-known problem that they reach too many voters who are politically engaged and pay a lot of attention to the news. Those voters tend to be more extreme than the population overall. To try to reduce this effect, The Times/Siena Poll makes extensive efforts to reach voters who are less likely to respond to polls. This drives up cost, but we think it’s worth it. For example: In 2024, we saw early on that Trump was building a lead with these voters.
Assessing other pollsters
Tell me about the major players in election polling data providers. Are the best data sources public or proprietary? — Jack McCumber, Seattle
Ruth Igielnik answers:
One of the best ways to assess whether a poll is trustworthy is how much information its pollsters give about how it was conducted. This methodological transparency has historically been correlated to survey accuracy over time. What does this mean in practice? Pollsters should tell you where they got their list of people to interview, how they contacted those people, exactly what questions they asked and in what order, when the poll was conducted, how they statistically adjusted the data afterward and who paid for the research. That may seem like a lot, but it is actually the bare minimum.
Along with transparency, another way to assess trustworthiness is to look for polls with a history of accuracy in previous elections. On our poll collections pages, we identify high-quality pollsters with a history of accuracy as “select pollsters” (indicated with a black diamond).
Plenty of pollsters who do not meet the “select pollster” designation can also be high quality. We include in our tracker all public polls whose validity can be evaluated by our polling collection team.
In addition to public pollsters, private campaign pollsters also sometimes release polling data, and many of them produce high-quality data, since their role is to inform politicians about how voters feel and help campaigns make informed decisions based on the state of the race.
Readers should be cautious, however, when they see a poll conducted by a group affiliated with a candidate or a partisan cause. Campaigns sometimes selectively release polls to tell a favorable story about their candidate or push a narrative. Polls that were conducted by or for partisan organizations are labeled on our poll tracking pages.
Readers should also exercise caution when new or untested polls are released. The barrier to entry to creating polls has never been lower, and this year there have been a small number of fabricated “polls,” making it harder for people to assess what is trustworthy.
Reporting results on Election Day
How do you determine when to declare a candidate the winner? — Meg McKeon-Baker, Queens
Lindsey Rogers Cook, a deputy editor on Election Analytics, answers:
We rely on The Associated Press for a majority of our race calls. It employs a team of analysts, researchers and race callers who have a deep understanding of the states where they declare winners. In a small number of high-profile contests, The Times independently evaluates A.P. race calls before making a projection. Both The Times and The A.P. will make a race call only when there is no possible path to victory for the trailing candidate.
Before we make or verify a race call, we want to be certain that the candidate will win that state or race. To determine this, we scrutinize where votes may remain to be counted in a given contest and what the likely makeup of those votes is. We are helped by pre-election estimates, statistical models, analysis of voting results and reporting from the county on remaining ballots.
Do you conduct your own exit polls on Election Day? Where? Do you as reporters know the results of those exit polls but hold that knowledge until polls close? — Gillette Hall, Santa Barbara, Calif.
Nate Cohn, The Times’s chief political analyst, answers:
We do not run exit polls, which have noted issues with accuracy. This year we launched Times Voter Trends, which will publish shortly after the election and break down estimates for the results of U.S. House and Senate elections by age, race, education and gender for every state in the country. To create these estimates, we will attempt to recontact the roughly 30,000 voters we interviewed this year for Times/Siena polls to ask them whether they voted and who they voted for. We will combine that information with actual election results at the precinct and county level to estimate how demographic groups voted.
Reporting on the races
How are we to look at these polls now and feel any confidence that they are accurately capturing the national environment and the reality of where the vote will be? — Jay Marcus, Los Angeles
Nestor Ramos, National editor, answers:
Polling is an incredibly useful tool to help understand the state of a race, and you can read about how our polling team does much of that work in this Q&A. But while it can tell you how people say they would vote, it’s just as important to understand the factors that are driving their decisions.
To help answer some of the questions that polls often can’t, we rely on on-the-ground reporting from all over the country. We have reporters in battleground states talking to voters every day to try to gain insight into how they are feeling about politics, policy and everything else that is motivating them to vote (or, as is the case for so many Americans, to stay home).
In just the last few weeks, Sabrina Tavernise has visited the key battleground state of Michigan to talk to voters and found person after person who had lost faith in President Trump. And in Pennsylvania and elsewhere, she found a bipartisan swath of voters who had lost trust in not just particular politicians but the entire system.
What you find when you talk to enough voters is often pretty nuanced. Sometimes opinions break cleanly along political lines. But often you find people who are earnestly wrestling with how big political questions affect their own lives. What will make their communities safer? Which candidate or party do they think will help them afford the life they want?
How many reporters are physically reporting from outside the New York metro area? Are reporters continually on the ground in states like Texas, Georgia, North Carolina, Alaska, Maine? If not, why and how does this affect their ability to report? — Susan Cauldwell, Middletown, Del.
Michael Cooper, Politics editor, answers:
There’s no substitute for on-the-ground reporting when it comes to trying to understand and explain the contours of campaigns and the issues animating voters. We have reporters living full-time in several of those states, some assigned to the Politics desk and others to the National desk, and they are invaluable as we cover the midterms. The Times recently expanded its team in Texas, which has been hugely helpful covering its Senate contest, which has been one of the most interesting races this year. And in states where we do not have reporters permanently based, we make frequent reporting trips. In the past couple of months, for instance, three different Times reporters have traveled to Alaska to cover various aspects of its crucial Senate race, flying all over the state and even chartering a boat at one point.
What policies does The Times have in terms of reporters (and other employees) contributing to (monetary or in-kind donations) or supporting (volunteering, bumper stickers, yard signs, etc.) political candidates and campaigns? — Eric J. Francis, Rochester, N.H.
Susan Wessling, Standards editor, answers:
The Times has clear policies, laid out in our ethics handbook, against newsroom staff members donating to a campaign, volunteering on behalf of a candidate, wearing campaign buttons or displaying any kind of political partisanship while on the job.
How do you handle “no comment” situations — it appears that The Times is facing a wall of government officials and spokespeople whose only job is to refuse to answer questions. — Andrew Finlayson, Franklin, Tenn.
Mike Abrams, deputy editor for Trust, answers:
What you see in our coverage is definitely what our reporters experience in Washington, on Wall Street and well beyond. But it’s essential that we seek comment as a matter of fairness and accuracy.
Sometimes those we reach out to ignore us. Sometimes they post our questions on social media to sway sentiment before we’ve published a word. And sometimes they just say “no comment.”
Even if it’s unsatisfying to read that kind of response, we believe it’s important to share the results of our reporting for the sake of fairness, accuracy and transparency. Want to learn more? Colleagues on our Business desk recently explored this topic for a Times Insider piece.
Media influence on the races
Media coverage, especially from a reputable brand with vast viewership like The Times, can have a very real impact on a race. As we’ve seen the 2026 battleground expand, how does The Times contend with the impact that its focus can have? — Micah Tjader, Madison, Wis.
Michael Cooper answers:
Political newcomers are often newsworthy, and during this cycle we have focused on some of the more interesting ones, mostly in battleground states. We have written about Republican athletes and sports figures running for office, Democrats looking for new types of candidates to compete in Republican areas, a Republican YouTuber who posts about guns who is running for one Texas House seat and a Democratic Tejano star who is running for another. But in key races where a newcomer is challenging an incumbent, we try to explore the records and backgrounds of both the challenger and the incumbent, as we have done in Maine, where we looked at Troy Jackson, the Democratic nominee, and Senator Susan Collins, the Republican incumbent.
Answers were contributed by Mike Abrams, Nate Cohn, Lindsey Rogers Cook, Michael Cooper, William P. Davis, Ruth Igielnik, Nestor Ramos, Caroline Soler and Susan Wessling.
The post Answers to Your Questions About Our Midterm Election Coverage appeared first on New York Times.
