President Trump departs the White House on Feb. 16 for a trip to his private Mar-a-Lago resort in Florida. (AP)

According to no less an authority than President Trump, “No politician in history … has been treated worse or more unfairly” by the media than Donald Trump. If public opinion was a direct reflection of news coverage, Trump’s popularity would be cratering by now. Instead, a recent rebound has left his approval rating just four points below where it was when he entered the White House.

Even more remarkably, people’s impressions of Trump’s personal qualities have been virtually unaffected by his precedent-smashing campaign and presidency. A November 2017 YouGov survey invited respondents to rate Trump as “intelligent,” “a strong leader,” “knowledgeable,” “inspiring” and “moral.” The results were barely distinguishable from responses to the same questions from the same 2,000 people in July 2016, just before Trump won the Republican presidential nomination. None of the average trait ratings changed by as much as five points on a 100-point scale, and the overall erosion in impressions of Trump’s character amounted to less than two points.

The July 2016 survey was conducted before the release of Trump’s notorious “Access Hollywood” tape; before Trump basked in his supporters’ chants of “Lock her up!” during and after the campaign; before President Trump fired FBI Director James B. Comey in the midst of his investigation into possible collusion between the Trump campaign and Russia; and before Trump told Puerto Ricans that Hurricane Maria was not a “real catastrophe” and tossed paper towels to the victims.

How could all of this (and much more) have had so little impact on people’s views of Trump?

One obvious stabilizing factor is the tendency of partisans on both sides to see what they want to see. While Republicans, independents and Democrats provided rather similar relative ratings of Trump (lowest for “moral” and “inspiring,” higher for “intelligent” and “strong leader”) and even shifted their ratings in roughly similar ways between surveys (with relative gains for “moral” and losses for “intelligent” and “strong leader”), the gulf in absolute ratings between the three groups was substantial. Republicans’ average ratings on the 100-point scale in November 2017 ranged from about 60 to 72, independents’ from about 36 to 47, and Democrats’ from about 14 to 21. (These tabulations are based on partisan identification as measured in 2015, before Trump emerged as the likely Republican nominee, so partisanship is not simply a proxy for Trump support.)

What is most surprising here is not the mild increase in partisan polarization of impressions of Trump (from an average difference of 44 points between Republicans and Democrats in July 2016 to almost 50 points in November 2017), but the relative stability of ratings within each partisan group and across all five dimensions of evaluation. The correlations between individual respondents’ ratings on each dimension in the two surveys ranged from 0.69 to 0.74, and the correlation between average ratings across all five dimensions was even higher, 0.79. Moreover, Americans seem to be gravitating toward increasingly undifferentiated impressions of Trump, producing stronger correlations in individuals’ ratings across the five distinct dimensions; the average correlation between pairs of ratings increased from 0.82 in July 2016 to 0.88 in November 2017.

The impressive stability of ratings of Trump’s character for individuals and partisan groups across trait dimensions over 16 tempestuous months seems hard to square with any straightforward account of political learning. But, strange as it may seem, that stability may be an ironic reflection of the media’s energetic policing of Trump’s malfeasance.

The news media have set a new standard for negative coverage

The president’s complaints about negative media coverage are not just self-aggrieved rantings. Studies by the Pew Research Center, Harvard’s Shorenstein Center and the conservative Media Research Center have all found that news coverage of Trump “set a new standard for negativity,” as the Harvard report put it. Pew researchers calculated that media coverage of Trump’s first two months in office was 62 percent negative and just 5 percent positive. By comparison, Barack Obama’s coverage in the first months of his presidency was 20 percent negative and 42 percent positive.

Many defenders of the media (and opponents of the president) are unruffled by the unprecedented negativity of Trump’s coverage. As Post media columnist Margaret Sullivan argued, “If a president is doing a rotten job, it’s the duty of the press to report how and why he’s doing a rotten job.”

But as any parent knows, a steady diet of criticism — however well-deserved — risks being tuned out, especially if it is delivered in a hectoring, combative or dismissive key. Legendary journalist Bob Woodward has criticized “a kind of self-righteousness and smugness” in news coverage of Trump. “When we reported on Nixon, it was obviously a very different era but we did not adopt a tone of ridicule.”

Media critic Howard Kurtz blames both the messengers and the reality they have to report: “Some stories are legitimate, some are not, and others are generated by the president’s own falsehoods and exaggerations. But the mainstream media, subconsciously at first, has lurched into the opposition camp.” Kurtz adds, “Donald Trump is staking his presidency … on nothing less than destroying the credibility of the news media; and the media are determined to do the same to him.”

In light of these developments, it should not be surprising that journalists themselves have become prominent protagonists in the partisan battle, continuing a long trend of declining confidence in the press, especially among Republicans. In the November 2017 YouGov survey, Democrats and Democratic “leaners” (as measured in 2015) gave journalists an average rating of 6.3 on a zero-to-10 scale; the corresponding average rating among Republicans and Republican “leaners” was 3.6. That 2.7-point gap exceeded the partisan difference in ratings of such familiar right-wing bugbears as immigrants (1.3 points), Muslims (1.8 points), gays and lesbians (1.9 points), and college professors (2.5 points).

Even more worryingly, the negativity of news coverage of Trump seems to be producing “outrage fatigue” not only among the president’s supporters but also among the shrinking sliver of the public that has so far resisted partisan polarization. The 14 percent of YouGov survey respondents who were “pure” independents in 2015 gave journalists a tepid average rating of 4.2 in November 2017 — a good deal closer to the Republicans’ 3.6 than to the Democrats’ 6.3.

If even these hardy surviving neutral observers are souring on the press and mostly tuning out what it has to say about Trump, the prospects for significant political learning, much less civic renewal, in the months and years ahead seem bleak indeed.

Larry M. Bartels holds the May Werthan Shayne Chair of Public Policy and Social Science at Vanderbilt University. His books include “Democracy for Realists: Why Elections Do Not Produce Responsive Government” with Christopher H. Achen, (Princeton University Press, 2016) and “Unequal Democracy: The Political Economy of the New Gilded Age,” 2nd edition (Princeton University Press, 2016).