Close Mobile Menu

Changing Minds about Changing Minds: How Results Once Faked Could Actually Be True

June 20, 2016
by Brendan Buhler

It may be time to change our minds about the impossibility of changing people’s minds. Again.

Besides your puny life experiences and all of literature and history, there’s mountains of evidence from the social sciences to suggest that it’s nearly impossible to change someone’s mind. It almost doesn’t matter what the subject is—whether you’re trying to persuade someone that vaccines are safe or that there are fewer immigrants than they think there are. You can try to change people’s minds or you can await the heat death of the universe—you’re as likely to live to see either of those events.

A quick recap of social sciences research on the topic: People are stubborn jerks. There are a lot of different terms for this with varying shades of nuance and meaning: motivated reasoning, cognitive dissonance, boomerang effect, Semmelweis reflex, pride and prejudice. Whatever you call this, we’re all familiar with it. We’ve all learned to hold our tongues at holiday dinners and not argue with Uncle Fred even when there is actual, factual evidence that he is wrong, because he will just yell and then no one will enjoy their pie. Call the problem Uncle Fred’s Obstinateness, or UFO.

Opening hearts could open minds. It was empi­ri­cal and squishy and just so gosh-darned inspir­ing.

(It’s UFO, incidentally, that drives major elections. Since you can’t change anybody’s mind, the reasoning goes, you want to rile up the voters who already agree with you and depress those who don’t, so they’ll stay home on election day. This is why presidential campaigns tend to feel like trench warfare without the joie de vivre.)

Given the UFO problem, it was exciting news in 2014 when a UCLA Ph.D. student named Michael LaCour published a paper in that most respectable of journals, Science, and said he’d found a way to change people’s minds. He showed that if you go door-to-door in Los Angeles to have short personal conversations with people, you could convince them to support marriage equality. They weren’t just humoring their interlocutor, either; you could check back with them months later, and they were still convinced. LaCour’s study even suggested a plausible mechanism: The conversations only worked when the canvassers were gays or lesbians, suggesting that by putting a face on the issue, people’s empathy could be engaged. Opening hearts could open minds. It was empirical and squishy and just so gosh-darned inspiring. Ira Glass did a big story on it on radio’s This American Life.

UC Berkeley Ph.D. students Joshua Kalla and David Broockman were so impressed with this breakthrough that they got a grant and set out to expand upon the experiment. They tried to replicate LaCour’s first step, signing people up for an online survey—but the response rate was significantly lower than LaCour’s. What were they doing wrong, they wondered. They looked into the methodology a little more; they called around. It turns out, LaCour’s data was so good for a very simple reason: He faked it.

It was literally too good to be true. LaCour’s big-name coauthor was embarrassed. Poor Ira Glass had to retract the story. And the rest of us, well, we had to go back to living under the shadow of UFO.

What about Kalla and Broockman? They won a prize from the Berkeley Initiative for Transparency in the Social Sciences and were exemplars of scientific rigor. Sure. Great. But they also still had a $95,000 grant and an experiment to run. So they ran it. And they found that it is possible to change people’s minds. What had been faked could actually be true.

The study Kalla and Broockman ran was conducted in conservative neighborhoods of Miami, and aimed to see if people could change their minds about transgender rights. Though it turned out people could change their minds, there were, reassuringly, some differences between their real study and LaCour’s fake one. First, support for transgender rights didn’t go up quite as much—10 percent instead of LaCour’s 14 percent.

Support for trans­gender rights among those can­vassed went up by the same amount that nation­wide accep­tance of gay rights did in the 14 years between 1998 and 2012.

Does that mean there are 10 percent more supporters of transgender rights at the end of the study? Not exactly—participants are not being asked one yes/no question, but rather a collection of questions to measure their knowledge and attitudes. “You can almost think of it as an SAT of transgender rights,” Kalla says. “Just like if I wanted to know your mathematical aptitude, I wouldn’t just ask you one question, I’d ask you 20 questions and combine them together.” A good frame of reference for the results, Kalla says, is that the support for transgender rights among those canvassed went up by the same amount that nationwide acceptance of gay rights did in the 14 years between 1998 and 2012.

More interesting is that it didn’t matter whether the people knocking on doors and talking to people about transgender rights were themselves transgender. We don’t know why this should be so—the empathy mechanism suggested by the phony LaCour results rests on a rich and unfaked body of research—but it’s a fascinating question.

And more questions are just what Kalla and Broockman would like to see. They took great care to describe their methods and mathematical models so that others can repeat the experiment and design new ones to tease apart how these long interviews change minds and whether the technique can be applied to other issues besides transgender rights. Maybe the most exciting thing about their real experiment is that, working with Berkeley political science professor Jasjeet Sekhon, they’ve come up with a way to run these studies at a fraction of what they would otherwise cost.

Kalla and Broockman even ran a small experiment to see whether persuasive canvassing could change people’s views on abortion, whether there was perhaps hope that even on the most divisive of topics we could find a way to stop talking past each other. And?

“It had no effect on changing attitudes toward abortion,” Kalla says.

Or, as the paper says, “This study estimated a precise zero.” 

Brendan Buhler doesn’t have an actual uncle named Fred.

Share this article