A myth is a nearly impossible thing to fully dislodge from the hearts and minds of the public.

They persist unfettered by accuracy or logic, like a meme equivalent of the common cold. It doesn’t take everyone believing in any particular myth or rumor to wreak havoc, only enough of us.  Though some myths and urban legends are relatively harmless, like sewer alligators in New York, others have warped the landscape of our society, leaving us to flail widely against nonexistent bogeymen. Rates of the almost-vanquished disease measles have increased on the word of a since-disbarred doctor, and sorely needed climate change legislation has stalled as politicians recite debunked arguments on the Congress floor.

These examples and more have led social scientists to try to understand how and why we continue to believe in the unbelievable; and in the case of MIT researcher Adam Berinsky, to find out just how to convince us otherwise. His latest study, “Rumors, Truths, and Reality: A Study of Political Misinformation,” set to be published in the British Journal of Political Science, has come to some rather unconventional conclusions. Namely that repeating a myth, even if in the course of debunking it, can only reinforce our belief in it. And that the best way to steer people towards rejecting a politically motivated myth is by recruiting a credible but idealogically opposed source to do it for you.

Using the real-life myth that the Affordable Care Act (ACA) contained provisions for so-called “death panels” that would deny health care to seniors depending on their level of productivity, Berinsky devised two experiments in 2010. In the first, he presented subjects news stories with different types of corrections of the myth, alongside articles that simply repeated the myth with no corrections.  He then polled them to find out how many of each group rejected the myth immediately after reading their article, following up with them a week later to ask again.

One version of the article debunked the ACA myth by citing nonpartisan sources such as the American Medical Association; another debunked it by accurately quoting a Democrat who drafted the legislation; while yet another quoted a Republican who drafted it. People who were exposed to an article which quoted a Republican rejected the myth to a greater degree than any of the other groups, though as Berinsky predicted, these levels dropped between the two surveys. The effect was particularly significant among those Berinsky identified as attentive participants, who rejected the ACA myth at a 69 percent clip when reading a Republican-debunked article as opposed to a 60 percent rejection rate for those who read articles citing a nonpartisan or Democratic source.

“Most importantly, the Republican correction is the most powerful treatment for both Republican and Democratic identifiers,” Berinsky wrote in his paper, ”[I]t is the informational content of the identity of the politician making the correction that matters.”

Berinsky speculates that seeing a Republican debunk a myth that might be advantageous to their political party was jarring enough for the participants to take stock and reevaluate their own position. “In particular, corrections acquire credibility when politicians make statements that run counter to their personal and political interests,” he wrote.

These considerations matter because, despite what we might think, the degree to which any of us form our beliefs is heavily influenced by our past experiences and social affiliations, myths included.  That’s why, as Berinsky reports, debunking a myth that touches upon our personal convictions can backfire and leave us on the defensive, more eager to believe the myth than ever before. “For instance, conservatives who received information that Iraq did not possess weapons of mass destruction were more likely to believe Iraq had those weapons than were respondents who did not receive the correct information,” he wrote.

The other important finding of Berinsky’s study is that reminding people of a myth can also backfire by reinforcing the idea in their minds, making it easier to misremember it as true. Sure enough, in the second experiment, Berinsky’s subjects were more likely to accept the “death panel” myth when asked to recall the entirety of the article they read, including the myth, as opposed to recalling an irrelevant detail of the story. This held true even for those who were fed both the myth and its correction. “Simply asking subjects to repeat the rumor to themselves without any indication that it is true increases their willingness to believe the existence of death panels, even weeks after they read the initial story,” he wrote.

Despite his somewhat disheartening findings, Berinsky feels that his results can help guide us towards effectively plugging up the rumor mill. While the repetition of a myth can spread its belief, Berinsky suspects that corrections can work that way too. And rather than pretending that people can simply be swayed by a rote recitation of facts by objective sources, what Berinsky’s study indicates is that our identity shapes why we believe what we believe.

By playing to that sense of identity and finding those who are willing to stand for accuracy, even when they might disagree with us philosophically, we can create a more honest engagement with the public.