The era of Donald Trump has brought with it what one might view as either a golden moment or a dark age for fact-checking. The president’s extremely loose regard for the truth, even for a politician, has produced a surfeit of fresh grist to verify—or more often, debunk. Yet it doesn’t seem to make a great deal of difference. Trump’s approval rating is in the basement, but it’s his inability to get anything done and his legal troubles that seem to be hurting him; his dishonesty was already manifest during the campaign. If the bleaker view is right, then truth is a casualty of the age, and stories like David Leonhardt and Stuart A. Thompson’s exhaustive inventory of Trump’s lies (their word) are little more than exercises in liberal and anti-Trump catharsis.

If you’re in the latter camp, then a new political-science paper has good news and bad news. Here’s the good news: Trump voters do not, in fact, seem impervious to truth. Present them with a falsehood from their man and they’ll acknowledge he was wrong. But that doesn’t have much effect on their support for Trump. As the authors put it, “Individuals may be willing to change their minds about the facts, but we do not observe changes in the candidate whom they support.” They know he’s wrong, and they don’t care (that much).

The study comes from Dartmouth’s Brendan Nyhan and Jason Riefler of the University of Exeter, who have been behind some of the most important work on the impact of corrections and fact-checking in recent years, as well as definite-article enthusiasts Ethan Porter of the George Washington University and Thomas J. Wood of the Ohio State University.

The quartet looked at a pair of claims that Trump made while he was the GOP nominee for president in 2016. In one case, during the Republican National Convention in Cleveland, he claimed that crime was up sharply, which was false. In another, during the first presidential debate, he insisted that Ohio and Michigan were hemorrhaging jobs, when in fact both states had unemployment levels below the national average.

Previous research from Nyhan and Riefler has found that corrections can sometimes actually have a “backfire effect”: When confronted with contrary information, ideologically motivated respondents sometimes simply dig in further on the initial, incorrect view of a fact. One might assume, then, that Trump supporters who were faced with Trump’s untruths would reject the correction. But that’s not what happened. They conceded the factual discrepancy, but continued to support Trump anyway.

“Though partisan loyalties are often inflamed during campaigns, corrections can still bring people’s beliefs more in line with the facts, even if they support the candidate being corrected or the correction is disparaged by a co-partisan,” the authors write.

As the authors note, the study is limited in scope, but it’s still a useful aid to thinking about Trump and his dissembling. The fact that he is able to “get away” with such frequent dishonesty has perplexed many observers. (“Getting away” is in the eye of the beholder; Trump’s presidency has encountered a variety of self-constructed obstacles.) Some uncharitable observers have attributed this to naive unsophistication among Trump voters brainwashed by media outlets and incapable of telling fact from fiction; other, more complex interpretations have situated Trump as an improbable apotheosis of the post-modern critique of empirical fact.

If this study is right, though, neither of those views is accurate. Trump supporters can assess the evidence, and they haven’t thrown out truth. It’s just not the operative factor in their choice.

The question is whether Trump is unusual in this regard. All politicians bend the truth sometimes. But some politicians build their careers on technocracy, detail, and competency. Take Hillary Clinton, for example, who presented herself not as a transformative leader but as a manager. Clinton’s defenders complained bitterly that she suffered while Trump got away with repeated untruths. Politifact’s files show that, in fact, Trump is unusually untruthful, with 69 percent of his ranked statements judged to be mostly false, false, or “pants on fire.” A full half of Clinton’s statements, in contrast, were rated true or mostly true.

But it could be that Clinton was punished more harshly for her equivocation about her email server, for example, because competency was part of her “brand.” Trump, however, never made much pretense about fact; he bragged as far back as The Art of the Deal that he used blatant exaggeration as a selling tactic, and many of his claims during the campaign didn’t even have a flimsy basis in fact; they were fabricated out of whole cloth. So perhaps Trump voters understood that factual fidelity was not what they were getting from him, and didn’t feel upset to learn he wasn’t telling the whole truth.

This dovetails with what I heard from Trump voters during the campaign. Frequently, they’d say they wished Trump would choose his words more carefully, exaggerate less, or back away from his more radical policy positions. Ultimately, though, their support was based on image and tribal identity. They felt that Trump was speaking bluntly and fighting for them in a way other politicians did not. Salena Zito made a similar observation in The Atlantic, from which this new paper adapts its title.

Trump’s coalition was also, as often noted, far more ideologically diverse than the traditional GOP base. As a candidate, Trump promised to protect entitlements and rip up free-trade deals, reversing long-standing Republican stances. Nyhan and Riefler’s previous work on the backfire effect noted that ideology motivated some voters to reject corrections on matters of fact, but Trump’s voters were perhaps less ideological than the average past Republican voter.

It might be interesting, therefore, to see what effect fact-checks would have on supporters’ perceptions of a politician who combined a self-image of factual wonkiness with charismatic tribal appeal. Such a politician exists, and he was Trump’s predecessor in the Oval Office. Unfortunately, it’s too late to use the same method as this paper to assess how fact-checks influenced the views of Barack Obama supporters. The 44th president was, using Politifact’s ratings, slightly less honest than Hillary Clinton (look it up!) but significantly more accurate than Trump. Obama’s major run-in with the fact-checkers was a consequential one: His promise that “If you like your health-care plan, you can keep it” was deemed the 2013 Politifact Lie of the Year, and it seems to have badly hurt the Affordable Care Act, which didn’t rebound until Trump and his allies approached repealing it.

The value of fact-checks is not in their ability to lift up especially honest candidates or to tear down dishonest ones. Assessing what politicians say and providing that information to the public is an absolute good, regardless of its impact. But this paper helps provide a slightly clearer understanding of what impact it has on the most important politician of the moment—and, more importantly, his supporters.