Bryan Woolston / Reuters

There are many reasons President Donald Trump might lose reelection in 2020. He is deeply unpopular. Most Americans abhor his bigotry. His administration has been plagued by all manner of scandals. He has failed to live up to his many grandiloquent promises. The country may be sliding into a recession.

Put all of this together, and it’s easy to imagine Democrats riding a big blue wave to the White House next year. But I fear that it is somewhat more likely that Trump will be able to declare victory on November 3, 2020.

Trump’s approval ratings are the most commonly used metric for how likely he is to win reelection. At first sight, they hold a lot of comfort for the president’s opponents. According to FiveThirtyEight’s tracker, for example, more Americans believe he is doing a poor job as president than believe he is doing a good one: About 54 percent disapprove of his performance. Only 42 percent approve of it.

Not only is Trump very unpopular, but his unpopularity is also unlikely to reverse anytime soon. After all, his approval ratings have continually been underwater since the second month of his presidency, and have fluctuated remarkably little since then.

But Trump’s persistent unpopularity is not nearly as big a bar to reelection as many assume. It’s striking, for example, that Trump’s approval ratings are, at this point, very similar to those of two recent presidents who went on to win reelection by resounding margins. While 42 percent approve of Trump’s job performance, just 43 percent approved of both Barack Obama and Ronald Reagan at the same stage in their first terms.

What’s more, Trump actually appears to be more popular today than he was on the day he beat Hillary Clinton. We can get a sense of how his standing with the public has evolved since the 2016 campaign by looking at his personal favorability ratings. An average of 41 percent of Americans now say they have a good impression of him, while an average of 55 percent say they have a bad impression, for a negative balance of 14 percent. In the last polls taken before the 2016 election, an average of 38 percent of Americans saw Trump favorably, and an average of 59 percent unfavorably, for a negative balance of 21 percent.

Since an election is a choice rather than a referendum, it is misleading to focus primarily on an incumbent’s approval ratings. In 2016, Trump was elected despite being deeply unpopular for the simple reason that his opponent was also deeply unpopular. For Trump to lose his bid for reelection in 2020, voters don’t just need to dislike him; they need to dislike somebody else less. Is that likely to be the case?

The obvious way to gauge how Trump fares compared with his competitors is to ask Americans whom they intend to vote for in a direct matchup. As of now, such general-election polls show a mixed picture. Joe Biden handily beats Trump. Bernie Sanders also tends to lead Trump, albeit by a considerably smaller margin. But all the other major candidates, including Elizabeth Warren, Kamala Harris, and Pete Buttigieg, tend to run head-to-head with the president: While they narrowly lead Trump in many polls, they trail him in many others.

As Nate Silver has shown, it would be a serious mistake to assume that head-to-head polls taken at this early stage are a reliable guide to the future. After all, polls taken at the end of the year preceding presidential elections have, on average, been about 11 points off the final tally.

It would also, however, be foolish to disregard head-to-head polls completely. As partisanship has deepened over the past decades, early polls have come to be more accurate. In the five elections since the turn of the millennium, they have been off by an average of 6 percent; in the past two elections, they have been off only by an average of 2 percent. While these polls cannot predict the future, they provide an important check on our intuitions.

What we know so far is perfectly compatible with Democrats winning a resounding victory or Trump securing reelection by a comfortable margin. Right now, public opinion gives, at best, a very small advantage in the national vote to Democrats. Since the electoral college is, as in 2016, more likely to favor Trump than his opponent, that is very cold comfort.

A lot can change—in either direction. But there is one final reason to think the president’s chances of reelection are better than meets the eye.

Trump is a known quantity. After three years in which Democrats have—for good reason—attacked him from every possible angle, it is difficult to imagine that they might suddenly succeed in changing how most Americans feel about him. What new angle of attack is supposed to turn against Trump voters who have so far stuck with him?

By contrast, so far Republicans have not had the need or the occasion to concentrate their attacks on any one of the 16 Democrats running for the party’s nomination. Once they do, they are likely to decrease the popularity of whoever ends up emerging as the victor.

This is especially true if the eventual Democratic nominee has come to national prominence only over the past few years—for instance, Kamala Harris or Pete Buttigieg. But the conservative attack machine might also affect opinions about candidates who have been in the public eye for much longer, such as Joe Biden or Bernie Sanders. Remember, when Clinton stepped down as secretary of state, in February 2013, nearly two-thirds of Americans had a favorable view of her. By the fall of 2016, when she was the Democratic Party’s nominee for the presidency, just over one-third of Americans retained a positive view.

Since 1945, nine presidents have sought a second term in office. Of these, six were reelected. Two of the remaining three, George H. W. Bush and Gerald Ford, had succeeded presidents of their own political party, so they were essentially seeking a third or fourth term in office. The only president to lose a bid for reelection after winning power from the opposite party was Jimmy Carter—and he was facing unusually heavy headwinds due to a combination of an economic crisis at home and national humiliation abroad.

Trump is, of course, an abnormal president. And so it is perfectly possible that he will, in the end, also prove abnormal in a more prosaic way—by losing his bid for reelection.

But what is eminently possible need not be likely.

We want to hear what you think about this article. Submit a letter to the editor or write to letters@theatlantic.com.