clock menu more-arrow no yes mobile

Filed under:

Distrust of fact-checking is not restricted to the right

Even fact-checkers are viewed through partisan lenses.

What to do about the “post-truth” era in American politics? Many have hoped that beefed-up fact-checking by reputable nonpartisan organizations (e.g., Fact Checker, PolitiFact, FactCheck.org, Snopes) would equip citizens with a tool to help them identify lies, thereby reducing politicians’ incentive to do so.

Putting aside questions about the consistency of their methods, does fact-checking work? As mass polarization deepens and media sources continue to balkanize, citizens who see bias in the traditional purveyors of information (mainstream media, academia, etc.) may often dismiss fact-checkers as similarly biased. And even when fact-checking succeeds at correcting misperceptions, it may not reduce supporters’ embrace of the liar. Such behavior would be consistent with social psychologists’ understanding of motivated reasoning.

One might guess that such imperviousness to professional fact-checking would be more or less exclusive to Republicans. After all, their longstanding distrust of media and academia is well known. However, as was on full display in social media last year and at the Democratic National Convention, insurgent progressives may have little more faith in “the establishment” these days than do the most ardent Trumpists.

Rejection of fact-checking from either the right or left might be most pronounced when its conclusions shed positive light on the insurgents’ pet boogeymen (or women). Enter Hillary Rodham Clinton, that rarest of politicos who was, in May 2016 at least, the preferred target of both red and blue anti-establishment guerrillas.

Taking advantage of these circumstances, in May of last year, just a couple of weeks before the California Democratic primary, we administered a randomized experiment to a representative sample of Californians using Sacramento State University’s CALSPEAKS panel (n=622). After measuring party identification and primary vote intention, we prompted all respondents with the statement below:

“Nonpartisan fact-checking organizations like PolitiFact rate controversial candidate statements for truthfulness.”

We also exposed one randomized half of the sample to the following statement and image:

“Each presidential candidate's current PolitiFact average truthfulness score is placed on the scale below.”

Respondents’ mean evaluation of candidates’ truthfulness.
Graphic created by authors

We created this visual by placing PolitiFact ratings as of May 2016 on a point scale and averaging them for each candidate. The image validated the growing mainstream narrative about Donald Trump’s disdain for facts, as well as Bernie Sanders’s tell-it-like-it-is reputation. But most notably, it contradicted conventional wisdom in many circles on both the left and right by indicating that Clinton’s statements had actually been more accurate, overall, than those of any other candidate running for president (though the difference between Clinton’s and Sanders’s ratings was not statistically significant).

We then asked all respondents to gauge 1) the extent to which they viewed Clinton, Sanders, and Trump as “honest,” and 2) the reliability of such fact-checkers. We wanted to see how exposure to the PolitiFact graphic might affect how people would respond to both questions, and whether those effects might differ depending on whether a respondent was a Republican (n=191), a Sanders supporter (n=154), a Clinton supporter (n=166), or an “undecided” Democrat/independent (n=111).

Perceptions of relative candidate honesty

Focusing first on the graphic’s effect on evaluations of Clinton’s honesty, relative to that of her two rivals, the charts below reveal that seeing the graphic increased Clinton supporters’ assessments of her honesty by about 9 percentage points relative to Trump’s, and by about 12 percentage points relative to Sanders’s. These differences are pretty small, but could in theory affect supporters’ enthusiasm and therefore turnout.

Citizen perception of candidate honesty: Clinton versus Trump.
Graph created by authors
Citizen perceptions of candidate honesty: Clinton versus Sanders.
Graph created by authors

Moreover, among potential primary voters who were at that time still undecided, exposure to the PolitiFact graphic seemed to increase mean appraisal of Clinton’s honesty by about 14 percentage points relative to Trump (p<.05), and 13 percentage points relative to Sanders (p<.01).

By contrast, and unsurprisingly, Republicans appear to have been unmoved by the PolitiFact graphic; exposure did not affect their evaluations of Clinton’s honesty (or Trump’s, for that matter).

More notably, the same was true of Sanders supporters (who in general were even more likely than Republicans to view Clinton as dishonest, regardless of whether they saw the graphic).

All told, fact-checking seems pretty toothless when it comes to persuading citizens to change their impressions of a political opponent — even when the opponent is in the same partisan family.

Perceptions of fact-checker reliability

What about perceptions of fact-checker reliability? Here the effects are much more striking. As the chart below reveals, when seeing a graphic implying that Clinton is more truthful than her reputation in some quarters suggests, the conclusion many of Clinton’s detractors drew was that it must be “fake news.” Specifically, among those who saw the graphic, Sanders supporters were about 13 percent less likely to view fact-checkers as reliable, and Republicans were about 35 percent less likely to do so.

Citizen assessment of fact-checkers’ reliability, by exposure to PolitiFact graphic (0 = unreliable; 1 = reliable).
Graph created by authors

Furthermore, and somewhat unexpectedly, undecided Democrats and independents also tended to react quite negatively to the graphic. They were about 33 percent less likely to view fact-checkers as reliable, compared with those who did not see the graphic.

The gist: fact-checking is no panacea

To conclude, the results of this experiment suggest that encountering surprises on PolitiFact’s Truth-o-Meter might increase supporters’ enthusiasm toward their candidate, and might even encourage undecided voters to take another look. It does not, however, appear to cause a positive reevaluation of candidates whom voters oppose.

Finally, whatever impact fact-checking might have on how citizens view candidates appears dwarfed by those citizens’ tendency to discredit fact-checking when they don’t like what they see — even among Democrats. This implies that the fact-checkers may have less and less influence over time, as more and more citizens encounter fact-checks that don’t jibe with their preconceived notions, until the whole exercise becomes pointless. And that should worry those of us who have hoped that fact-checking could help re-route the “post-truth” advance in American politics.

David C. Barker is a professor of government and director of the Center for Congressional and Presidential Studies at American University. Find him on Twitter @barkerccps.

Kim L. Nalder is a professor of government and director of the CALSPEAKS Public Opinion Research Center and the Project for an Informed Electorate at California State University Sacramento. Find her on Twitter @KimberlyNalder.

Danielle Joesten Martin is an assistant professor of government at California State University Sacramento. Find her on Twitter @djoestenmartin.

Sign up for the newsletter Sign up for Vox Recommends

Get curated picks of the best Vox journalism to read, watch, and listen to every week, from our editors.