Married men and veteran households were probably not the demographic groups that Democrats assumed would carry the party to victory over Donald Trump in the 2020 election.

But Joe Biden’s apparent strength among traditionally moderate or even conservative constituencies, and especially men, is emerging as one of the hallmarks of his victory, according to new data from Pew Research.

Trump won married men by just a 54-44% margin — a net 20-point decline from his 62-32% victory in 2016. He won veteran households by a similar 55-43% margin, down a net 14 points from his 61-35% victory.

In both cases, the size of Biden’s gains among these relatively conservative groups rivals Trump’s far more publicized surge among Latino voters. Each group represents a larger share of the electorate than Latinos, as well.

The Pew data, released today, is the latest and perhaps the last major batch of high-quality data on voter preference and turnout in the 2020 election, bringing analysts close to a final, if still imperfect, account of the outcome.

The data suggests that the progressive vision of winning a presidential election simply by mobilizing strong support from Democratic constituencies simply did not materialize for Biden. While many Democrats had hoped to overwhelm Trump with a surge in turnout among young and nonwhite voters, the new data confirms that neither candidate claimed a decisive advantage in the highest turnout election since 1900.

Instead, Trump enjoyed a turnout advantage fairly similar to his edge in 2016, when many Democrats blamed Hillary Clinton’s defeat on a failure to mobilize young and nonwhite voters. If anything, Trump enjoyed an even larger turnout edge while Biden lost ground among nearly every Democratic base constituency. Only his gains among moderate to conservative voting groups allowed him to prevail.

The Pew data represents the only large, traditional “gold standard” survey linked to voter registration files. The files reveal exactly who voted in the election, offering an authoritative evaluation of the role of turnout; but they become available only months after the election.

In previous cycles, the higher-quality data released months or years after the election has complicated or even overturned the narratives that emerge on election night. For this cycle, the Pew data — and other late analyses, like a study from the Democratic data firm Catalist — has largely confirmed what analysts gleaned from the vote tallies in the days after the election.

If anything, the newest data depicts a more pronounced version of the early analysis.

The Pew data, for instance, shows Trump faring even better among Latino voters than any previous estimate, with Biden winning the group by a 59-38% margin — a net 17-point decline from Hillary Clinton’s 66-28% victory in the same survey four years ago.

Trump’s breakthrough among Latino voters was the most extreme example of the broader inroads he made among Democratic constituencies. According to the data, Biden failed to improve his margins among virtually every voting group that backed Clinton in 2016, whether it was young voters, women, Black voters, unmarried voters or voters in urban areas. Often, Trump improved over his 2016 performance, even though he was largely seen as trying to appeal to his own base.

Higher turnout did not reshape the electorate to the favor of Democrats, either. In the aftermath of the 2016 election, many Democrats blamed Clinton’s defeat on low turnout and support from young and nonwhite voters. Many progressives even believed that mobilizing Democratic constituencies alone could oust the president, based in part on the assumption that Trump had all but maxed out his support among white, rural voters without a degree.

At the same time, Democrats supposed that higher turnout would draw more young and nonwhite voters to the polls, bolstering the party.

Overall, 73% of Trump’s supporters voted in the 2020 election compared with 68% of Biden’s supporters. In comparison, Trump’s supporters were only 2 percentage points more likely to vote than Clinton’s in 2016, according to the Pew data.

New voters, who did not participate in 2016 or 2018, split about evenly between Trump and Biden, with Biden winning 49% of new voters to 47% for Trump.

In the end, there was a far deeper well of support and enthusiasm for Trump than many progressives had imagined. An additional 13 million people voted for Trump in 2020 than in 2016. Voter records in states with party registration — like Pennsylvania, Florida, North Carolina, Nevada and Arizona — suggest that registered Republicans continued to turn out at a higher rate than registered Democrats, and in some cases even expanded their turnout advantage over the 2016 cycle.

Nationwide, Catalist found that the turnout among ‘historical’ Republican and Democratic voters both increased by 3 percentage points, leaving the basic turnout pattern of the 2016 election intact.

Whether the Democratic turnout should be considered strong or weak has been a matter of some consternation for Democrats, who are understandably reluctant to diminish the contributions their base made in ousting Trump. And of course, Biden absolutely could not have won the election if Democratic turnout did not rise to at least keep pace with that of Republicans.

Perhaps another Democrat would have mobilized voters more decisively. But the strong turnout for Trump implies that it would have been very challenging for any Democrat to win simply by outmuscling the other side.

Instead, Biden prevailed by making significant inroads among moderate or conservative constituencies.

Biden’s strength among these groups was not obvious on election night. His gains were largest in suburban areas, which are so heterogenous that it’s often hard to say exactly what kinds of voters might explain his inroads.

Biden’s weakness among Hispanic voters, in contrast, was obvious in overwhelmingly Hispanic areas like Miami-Dade County or the Rio Grande Valley.

According to Pew Research, Biden made larger gains among married men than any other demographic group analyzed in the survey. He won 44% of married men, up from 32% for Clinton in 2016. It’s an even larger surge for Biden than Pew showed Trump making among Latino voters, even though they do not stand out on the electoral map.

In a similar analysis, Catalist also showed that Biden made his largest inroads among married white men, though they showed smaller gains for Biden than Pew Research.

Biden also made significant, double-digit gains among white, non-Hispanic Catholics, a persuadable but somewhat conservative voting bloc. He won 16% of moderate to liberal Republicans, up from 9% for Clinton in 2016. And Biden gained among men, even while making no ground or, according to Pew, losing ground, among women. As a result, the gender gap was cut in half over the last four years, to 13 points from 26 points in 2016.

The shrunken gender gap in 2020 defies the preelection conventional wisdom and polling, which predicted that a record gender gap would propel Biden to victory. The Pew findings offer no insight into why the gender gap may have decreased; any number of interpretations are possible. In this case, it is possible that attitudes about Clinton may be a more important factor than attitudes about either of the 2020 election candidates.

This article originally appeared in The New York Times.

© 2021 The New York Times Company