Comparing the final NYT Siena poll to the 2016 and 2020 elections

The finale New York TimesThe Siena College poll for the 2024 presidential election released Sunday shows Vice President Kamala Harris with a slight lead over former President Donald Trump in all but a slight swing state.

For the survey, 7,879 potential voters in the seven battleground states were interviewed by telephone. Among respondents, the poll showed Harris leading by three points in Nevada, by two points in North Carolina and Wisconsin, by one point in Georgia, and had the advantage in a virtual tie in Pennsylvania. Trump had the advantage in a virtual tie in Michigan and a four-point lead in Arizona. Neither candidate had more than 50 percent support in any state, continuing to give the impression of a close contest with no clear advantage for either candidate.

The poll was conducted between October 24 and November 2, with a margin of error of plus or minus 3.5 points in each state.

However, Harris would have more positive things to offer than Trump. Analysis of the survey by The New York Times found that those who decided late broke for Harris: 55 percent of those who had recently decided to vote supported her, compared to 44 percent for Trump.

New York Times/Siena College is one of America's most trusted pollsters. Aggregator 538 ranks first on its list of 282 based on its historical track record and transparency. Analyst Nate Silver ranks the company among the top two companies and gives it an A+ grade.

Newsweek The Trump-Harris campaign emailed for comment.

What did the 2016 and 2020 polls say?

The 2016 election hurt pollsters overall, as the vast majority showed clear trends in favor of a clear lead for former US Secretary of State Hillary Clinton.

Many blamed then-FBI Director James Comey and his announcement days before the election that he would reopen the investigation into Clinton's emails for scuttling an otherwise sound lead and misleading the race.

On the way to election day Just had found Clinton to have 45.9 percent support, compared to 42.8 percent for Trump nationally.

Meanwhile, Siena College's national poll of battleground states showed Clinton with a seven-point lead in Pennsylvania, a seven-point lead in North Carolina and even Florida.

Pennsylvania was famous for Trump, giving him a victory by a margin of 0.72 points – just about 44,000 votes – as did North Carolina by 3.66 points and Florida by 1.2 points.

The 2016 backfire led many pollsters to adjust the way they weighted various demographics to better capture the “Trump effect” in the polls. However, these changes did not prove to be as effective as pollsters had hoped.

The 2020 Siena College Poll found Joe Biden with a nine-point lead over Trump among likely voters nationwide. Biden won by a margin of 4.5 points, only half as much as the poll showed.

In the 2020 battleground poll released on November 1 of this year, Siena College found Biden ahead by six points in Arizona, three points in Florida, six points in Pennsylvania and 11 points in Wisconsin.

In the end, Biden won Arizona, even though most expected a Trump victory, but he narrowly secured the victory by a margin of 0.3 points – just about 10,500 votes. Trump clearly won Florida by 3.5 points, and Biden retook Pennsylvania, but again only by a slight margin of about 1.16 points.

Biden also ended up winning Wisconsin, but by a much smaller margin of 0.63 points.

Can we trust the polls?

Polls have suffered a significant loss of reputation in the last two elections. Polls suggested Clinton would beat Trump in 2016 and that Biden had a larger advantage when he ultimately won in 2020.

“Although the polls are not so precise that one can be confident that the election will be close, one cannot assume that the polls will again be as badly flawed as they were in 2016 or 2020,” Cohn said , chief policy analyst for the Justwrote in an article published in Upshot on Friday.

Both previous polls have underestimated Trump's support, with many noting in 2016 that his supporters were less likely to express support for the highly controversial candidate.

Cohn highlighted “major methodological changes” to the survey data, noting that “many of the worst-performing pollsters in 2020 have either made major methodological changes or disappeared from the map.”

However, some of these changes, made to compensate for Trump's underrepresentation, may have instead led to Harris' underrepresentation, Cohn warned.

“It is hard to overstate how traumatic the 2016 and 2020 elections were for many pollsters,” he wrote. “For some, further underestimation of Mr. Trump could pose a major threat to their businesses and livelihoods. For the rest, their status and reputation are at stake.”

He added: “If they underestimate Mr. Trump for a third time in a row, how can you trust their polls again?” he continued. “It is much safer, whether in literal self-interest or purely psychological, to find a close race than to bet on a clear Harris victory.”