The specter that most haunts Democrats as we hurtle toward the 2020 election is that Donald J. Trump managed to get himself elected president in 2016 despite terrible, historically unprecedented (for a major-party presidential nominee) unpopularity, as measured by favorability ratios (Gallup had him at 36 percent favorable/61 percent unfavorable on election eve). To put it mildly, Trump overperformed his favorability numbers, even if you take into account Hillary Clinton’s unpopularity, which was strong, although not in Trump’s neighborhood (Gallup had her favorability ratio at 47/52). And if you don’t want to rely on Gallup, the official exit polls gave Trump a 38 percent approval rating among voters who actually showed up, compared to HRC’s 43 percent.
Now that Trump is POTUS, it’s his job-approval numbers that are the most relevant measure of his popularity. And those who have been staring at his approval ratings have to wonder: Are they good enough, or close to good enough, for another Trump win if (a) he drags his opponent down in public esteem as he dragged down HRC, and (b) he gets the benefit of an Electoral College advantage via more efficient distribution of his popular vote?
Kyle Kondik of the University of Virginia’s Center for Politics takes a long look at this question, and comes up with some interesting analysis. In 2016, he notes, Trump did particularly well among voters who didn’t like either major-party candidate:
Trump won 47%-30% among the 18% of the electorate who held unfavorable views of both Trump and Clinton.
That might have reflected some sort of gap in the intensity of disdain for the two candidates, but it’s equally if not more plausible that general disgruntlement with the status quo and a “time for a change” sentiment after eight years of Democratic control of the White House gave Trump the benefit of the doubt more often than not. In any event, Kondik notes more recent evidence that Trump may not overperform his popularity next year:
[I]n some early ballot tests, there is some indication that Trump is not only failing to pick up support from people who don’t like him, but in some instances he does not appear to be winning every poll respondent who approves of his job performance.
The Fox News poll released last week found that Trump’s approval rating among registered voters was 46%, but he only attracted between 39%-42% of the vote in matchups against the top-polling Democratic presidential contenders (Joe Biden, Kamala Harris, Elizabeth Warren, and Bernie Sanders — Biden usually does the best in these head-to-heads with Trump at this point). The NBC News/Wall Street Journal poll released in early July found Trump’s approval at 45% among registered voters, but his support in ballot tests against the four top Democrats was just 42%-44%. The ABC News/Washington Post survey conducted right after the first debates found Trump at 47% approval among registered voters, but Trump was only at 43% against Biden: against the three others, Trump was between 46%-48%, effectively equaling his approval.
It’s true that trial-heat general election polls this far from November of 2020 aren’t terribly reliable in terms of vote share, but the negative gap between Trump’s popularity and the percentage of voters who want him reelected could be meaningful.
Trump’s path to victory probably entails him either improving his approval rating so that it’s in the mid-to-high 40s as opposed to the mid-to-low 40s, or running ahead of his approval by capturing a small but significant number of voters who don’t approve of him. But if Trump is actually losing a small number of voters who approve of his job performance, he may have a very hard time cobbling together another Electoral College majority.
It’s worth repeating that a general improvement in Trump’s job-approval rating appears unlikely; his level of popularity has been amazingly stable over time, and seems relatively impervious to anything he says or does — or to objective conditions in the country. Looking at the FiveThirtyEight averages (which usefully adjust results for pollster quality and partisan bias), Trump’s average approval rating since March 1 has narrowly oscillated between a high of 42.9 percent and a low of 41.1 percent. Right now, after weeks of yelling and screaming by and about Trump in the wake of his racist attacks on members of Congress, it stands at 42.5 percent. If you prefer RealClearPolitics’ unadjusted and unweighted averages (which are generally a bit higher, mostly thanks to the clearly inflated Rasmussen numbers released each day), his approval ratings since March 1 have varied from a high of 45.1 percent to a low of 42.2 percent, and currently stand at 43.9 percent. There’s just not much sign of anything dynamic happening, and it’s not like general perceptions of this universally known and powerfully repellent or attractive (depending on your perceptions) personality changing as his presidential term chugs along its steadily noisy track.
Again, the X factor in comparing Trump’s likely vote in 2016 and in 2020 is incumbency, which we tend to think of as an asset even when it’s not. If there is any incumbency advantage, it’s likely baked right into Trump’s approval ratings. As Kondik points out, the relationship between presidential approval and reelection-year vote share is strong. Among presidents winning reelection, Obama exceeded his election-eve approval rating by one point in 2012 (as did Reagan in 1984), and George W. Bush exceeded his by two points in 2004. The only recent president who exceeded his approval ratings in vote share by more than two points was Poppy Bush in 1992, and he got beaten like a drum.
It’s worth thinking about this, too: For all the talk about Trump’s boffo economy, overall public assessments of the direction of the country remain sour. According to RealClearPolitics averages, the right track/wrong track ratio is currently at 38/57. An unpopular president leading an unhappy country could still get reelected, but the widespread, 2016-based belief that Trump is on some clear trajectory to a second term is mostly superstition.