A feature of modern presidential politics is a grim one: Voters are generally asked to choose between two candidates they might not like very much.

Sure, partisans are generally enthusiastic about their party’s nominee, but in 2016, given how unpopular both candidates were, a sizable number of voters who cast ballots said they liked neither Donald Trump nor Hillary Clinton. According to exit polling conducted that year, about 1 in 5 voters said they liked neither Trump nor Clinton, about nine times as many as said they liked both major-party candidates.

That became important, given how close the 2016 election ended up being. As it turns out, while people who liked Trump and didn’t like Clinton voted heavily for Trump (as you’d expect), the current president also had an edge among people who disliked both him and Clinton. He won those voters by 17 points nationally — and by margins in the closest states that were likely enough to hand him the electoral college victory he needed.*

(The sample size of those who liked both candidates was too small to break out, symbolic in its own right.)

One question coming into 2020 was whether Clinton’s relative unpopularity was a function specifically of her candidacy or whether the Democrats could avoid numbers like those above simply because they’d have a better-liked candidate. We learned this week that they probably wouldn’t: Two new polls show favorability numbers for leading Democrats at or near where Trump’s are right now.

A Quinnipiac University poll released this week, for example, shows Trump’s favorability at only 40 percent — but former vice president Joe Biden’s at 44 percent and Sen. Elizabeth Warren’s (D-Mass.) at 38 percent. Margins of error mean all those numbers are functionally equivalent, but that’s not good news for the Democrats.

When we reported that data, we noted that this didn’t necessarily mean Trump would fare as well with the most discontented voters. In 2016, after all, he was a relatively unknown quantity; perhaps people who didn’t like either person figured he was at least worth a shot. So will he enjoy the same benefit of the doubt next year?

Quinnipiac provided us with more detailed data, which answers that question. In short: No.

The split between those who like or don’t like the candidates is remarkably similar to 2016. Trump is the only candidate liked by 36 percent of respondents, same as the percentage of voters three years ago. Biden is the only one liked by 41 percent — the same percentage of exit poll respondents who said the same of Clinton. The group of like-neithers is slightly smaller, at 12 percent.

How do they vote? According to Quinnipiac, they prefer Biden by a wide margin. This is a relatively small group of people, so the margin of error is fairly large, but Biden’s margin is large enough that it doesn’t matter.

Note that only 10 percent say they’d vote third party, half the percentage of the same group that said that in 2016. Another 9 percent say they wouldn’t vote at all.

What we see when we compare those numbers to 2016 is a flip in sentiment among the cynical voters. That year, Trump’s support among the like-neithers was about the same as his overall percentage; it was Clinton whose numbers plunged. In Quinnipiac’s poll, Biden holds strong with that group, while Trump’s numbers drop off a cliff.

We’ll add here the necessary and accurate caveat that this is general election polling well in advance of the election. But it does provide a hint at an important question for next year’s election, how those voters who dislike everyone will vote. The answer? They appear to dislike Trump more.

*By the numbers, exit polls had Trump leading Clinton among those who didn’t like either candidate by 37 points in Wisconsin, 25 points in Pennsylvania and 21 points in Michigan. That voting bloc made up 22, 17 and 29 percent of the vote in those states, respectively.