Maybe Americans think Obama was particularly good as president at, say, giving eulogies, in a way Clinton would not have been. And, sure, at earlier stages in their careers, Clinton and Obama had different views on issues, most notably the Iraq War. But by the end of the 2016 campaign, they were basically aligned on everything. The gap between Obama-Trump (19 percentage points) and Clinton-Trump (3 percentage points) is really large, and I’m not sure it’s based on policy differences.

Second, Obama’s 19-point advantage over Trump is huge. Obama is not some hypothetical figure: He ran for president twice, and 46 percent (2008) and then 47 percent (2012) of the country voted for the other candidate. Trump is not John McCain or Mitt Romney, you might say, and that’s fair. But count me skeptical that an actual Obama-Trump head-to-head race would have produced such a blowout. My bet is that strong support from Republicans and Republican-leaning independents would have gotten Trump into at least the low 40s in such a contest. And, yes, I get that the poll asked who would be a better president, not who people would vote for. But the Clinton-Trump numbers seem much closer to how Americans actually act on their political preferences: On Election Day in 2016, the former secretary of state won 48 percent of the vote nationally, compared with Trump’s 46 percent. That roughly parallels who Americans think would be the better president.