Sen. Bernie Sanders’s (I-Vt.) victory in New Hampshire’s Democratic presidential primary on Tuesday night was the first clear result of the 2020 nominating contest. It was also the second time in a row in which Sanders has triumphed in the state, prompting obvious comparisons between the two contests.

Four years ago, Sanders crushed Hillary Clinton by more than 22 points. This year, he narrowly edged out former South Bend, Ind., mayor Pete Buttigieg, earning 25.8 percent of the vote (as of writing) to Buttigieg’s 24.5 percent.

This has prompted a slew of questions about the solidity of Sanders’s support and ancillary questions about his ability to turn out his base of support. A clear victory for his campaign has been muddied by his own success in 2016.

It’s worth answering the question: What happened?

One thing that happened, of course, is that Sanders was running against a much bigger field of candidates. He and Buttigieg combined to earn less than half of the total vote. Eight candidates earned at least 1 percent of the total votes cast. Four years ago, the only real contenders were Sanders and Clinton; now, Sanders had at least four other significant competitors: Buttigieg, Sens. Amy Klobuchar (D-Minn.) and Elizabeth Warren (D-Mass.) and former vice president Joe Biden.

In every township for which we had data Wednesday morning, Sanders had a lower percentage of the vote than he did four years ago. In 119 townships, he won both on Tuesday and four years before. In another 84 townships, he won in 2016 but trailed this time around.

In most of those 84 townships, the winner this time wasn’t Warren, who would seem to be the natural second choice for Sanders supporters. Instead, the winner was usually the more-moderate Buttigieg.

By itself this doesn’t tell us much. After all, it’s possible that in many places Warren’s being on the ballot might have siphoned off Sanders’s support and allowed Buttigieg — a more Clintonian candidate, politically — to move into the lead.

Exit polling suggests that the more significant factor was likely that the electorate itself was much friendlier to a Buttigieg/Klobuchar candidacy. Demographic groups that four years ago propelled Sanders to a massive win turned out less heavily in 2020, and Sanders’s margins with those groups were smaller.

Below you can see how exit polling captured the 2016 and 2020 electorates. (Data for 2020 are still preliminary.) We’ve broken out demographic groups into categories and scaled each group vertically depending on how much of the electorate the group constituted.

Looking at gender, you can see that the row indicating the choice of male voters is shorter in 2020 than it was in 2016, meaning that male voters made up less of the electorate on Tuesday than four years ago. Women made up more of the electorate this time around — and they preferred Buttigieg to Sanders by a small margin.

Across the demographic groups we broke out, you see a similar pattern. Sanders did worse with younger voters this time around, but it was still a key part of his base. Unfortunately, though, they made up less of the electorate. Older voters made up more of the electorate, and Sanders did significantly worse with that group.

Perhaps most significantly, the number of voters who identified as moderates was higher than four years ago, and Sanders did far worse with this group than he did then. There were far more options for moderate voters this time around, with Buttigieg, Klobuchar and Biden earning a bit more than half of the vote. The density of very liberal voters, with whom Sanders did well, was down.

On a percentage-point basis, Sanders fared worse in 2020 than he did in 2016 in each of the townships for which we have data. If we combine his support with Warren’s to form a sort of liberal-voting bloc, that combined vote exceeded Sanders’s 2016 support in only 23 townships.

If we combine the Buttigieg, Biden and Klobuchar support, they outperformed Clinton’s 2016 percentages in 174 townships for which we have data.

The question Sanders’s team will want to answer is why this electorate was so much more moderate. A decline in enthusiasm for his candidacy? Enthusiasm among moderates for the wider set of options on the ballot? An increase in turnout generally, which meant more moderates coming to the polls? Voters are afraid that Sanders can’t beat President Trump in November?

None of this detracts from the fact that Sanders again won the New Hampshire primary. It just suggests that similar electorates moving forward could mean similarly close fights.