clock menu more-arrow no yes mobile

Filed under:

These scientists can prove it's possible to reduce prejudice

Remember the same-sex marriage study that was faked? The theory's been retested, and the results are very promising.

Adapted from igor kisselev / Shutterstock
Brian Resnick is Vox’s science and health editor, and is the co-creator of Unexplainable, Vox's podcast about unanswered questions in science. Previously, Brian was a reporter at Vox and at National Journal.

It was the most optimistic, feel-good conclusion social science had produced in years. In 2014, political scientists, writing in the journal Science, reported that humans — stubborn and cliquish as we are — are capable of a profound change of mind.

The researchers had demonstrated that a 20-minute conversation with a gay canvasser could make a lasting impression on a voter's willingness to support same-sex marriage legislation. That may sound small, but it was amazing: The vast majority of social science research finds that persuading people to change their attitude is hard, if not impossible. Years of research into sensitive topics like politics and vaccinations had found that advocacy usually blows up in the advocate's face. Targets of persuasion tend to double down on their previously held position or forget the intervention almost immediately.

The Science paper was "almost like an existence proof that attitude change is possible," Don Green, a professor at Columbia University and the senior author on the study, told me last April. And it suggested something breathtaking: that it was possible to reduce prejudice over time. In the past, marriage equality activists, anti-abortion advocates, and others had failed to sway minds with costly advertising campaigns. Here was a compelling case that human-to-human interactions are what really matter.

Which is why what happened next was so disheartening.

In May 2015, it emerged that Michael J. LaCour, a grad student at the University of California Los Angeles who'd done the bulk of the work on the paper, had faked all the data and lied about his funding sources. When Green found out, he asked Science for a retraction and got one.

Shortly after the scandal broke, I called David Broockman, who was then finishing up his PhD studies at the University of California Berkeley. He, along with fellow grad student Josh Kalla and Yale political science professor Peter Aronow, had uncovered the fraud.

Broockman and Kalla were in the midst of setting up their own experiment to replicate the findings with transgender phobia in Miami. In trying to figure out how LaCour pulled it off, Kalla and Broockman found many strange inconsistencies in his data. When they probed, the data fell apart. (LaCour had adapted a preexisting dataset and manipulated it to show positive results for his study.)

But Kalla and Broockman were resolved to go forward with their own experiment. Broockman told me he didn't know what to expect, other than that this idea — that humans are capable of change — was one worthy of a retest.

Well, the results are in. Today, Broockman — now a professor at Stanford — and Kalla, still a graduate student at Berkeley, have published their findings in Science. And they're filled with even more optimism than the faked study two years ago.

The study is titled "Durably reducing transphobia: A field experiment on door-to-door canvassing," and it is the first large-scale, real-world experimental effort that shows lasting opinion change is possible. Broockman and Kalla showed that a 10-minute conversation with a pro-transgender canvasser can influence opinions for at least three months.

It's a rare story in science, a rare story anywhere: where young idealists not only uncover one of the greatest frauds in recent scientific history but then manage to validate the very idea that the fraudulent study asserted.

Putting an unconventional canvassing technique to the test

In a typical canvassing conversation, a person knocks on the door and spews statistics and facts to convince you to vote for a ballot measure. Those interactions are at best instantly forgettable and at worst incredibly annoying.

Broockman and Kalla were studying a different type of conversation, one developed in the Leadership LAB, a program of the Los Angeles LGBT Center in the wake of California's Proposition 8 that banned gay marriage. Frustrated by the loss on Prop 8, the LGBT Center's Dave Fleischer set out to talk to voters about why they decided against marriage equality. The conversations became the basis for a new technique.

The key difference between Fleischer's technique, sometimes called "deep canvassing," and the standard model is that Fleischer has voters do most of the talking.

"The key part of this is having people think back on their real, lived experience in an honest way," Fleischer tells me. "Everything we do is driven by that."

In talking about their own lives, the voters engage in what psychologists call "active processing." The idea is that people learn lessons more durably when they come to the conclusion themselves, not when someone "bitch-slaps you with a statistic," says Fleischer. Overall, it's a task designed to point out our common humanity, which then opens the door to reducing prejudice.

Fleischer has showed me several videos the LAB has shot of real encounters his canvassers have had with voters, and it's hard not to think he's onto something.

In one video shot in Miami, a young canvasser named Virginia approaches an older South American man named Gustavo. Virginia is tattooed up to the neck, and Gustavo is wearing a sleeveless undershirt tucked into khaki pants. Virginia is a gender nonconforming person who identifies as neither male nor female.

In the beginning of their conversation, Virginia asks Gustavo how likely he'd be to support transgender rights legislation. Gustavo says he wouldn't support it because he's worried about predatory men using the law as an opportunity to enter women's bathrooms.

Virginia asks why he feels that way.

"I'm from South America, and in South America we don't like fags," he tells her.

This next moment is crucial: Virginia doesn't jump on Gustavo for the slur, and instead says, "I'm gay," in a friendly manner. Gustavo doesn't recoil. Actually, he becomes more interested.

Gustavo and Virginia go on to discuss how much they love their partners, and how that love helps them overcome adversity. Gustavo tells Virginia that his wife is a disabled person. "God gave me the ability to love a disabled person," he says, and that taking care of one another is why love matters.

"That resonate a lots with me," Virginia responds. "For me, these laws, and including transgender people are about that. They're about how we treat one another."

Now that Gustavo is in a place where he's more open, Virginia asks him to imagine what the worst thing could happen if he used a bathroom with a transgender person. He admits he wouldn't be scared. Then comes the breakthrough.

"Listen, probably I was mistaken," he says of his original position on trans rights.

Virginia ask him again if he'd vote in favor of banning transgender discrimination. "In favor," he says.

This video is of an interaction between one of Fleischer's canvassers and a resident in Los Angeles, California. She too turns around completely on transgender rights.

"Deep canvassing" can inoculate people against prejudice

For their new paper in Science, Broockman and Kalla studied hundreds of conversations like the one between Gustavo and Virginia. It's the same basic technique LaCour and Green purported to study, but adapted slightly for transgender issues.

In Broockman and Kalla's study, the canvassers showed participants both sides of the transgender rights argument through videos. Then the canvassers asked them to recall a personal experience with judgment or prejudice. After participants spilled their guts, they were encouraged to think about how their personal experiences related to the experiences of transgender people.

Broockman and Kalla's paper finds these conversations not only move people to be supportive of trans people and trans-inclusive legislation — but that effect also persists three months out.

But what gives their results so much weight is the strength of the study design. It's similar to the "gold standard" randomized, placebo-control design that drug companies use to detect the effect of their medications.

Broockman and Kalla originally sent letters to 35,550 houses in the Miami area, asking individuals to participate in an online survey and several follow-ups for a small reward. That effort yielded 1,825 individual responses for a baseline opinion data.

Then they randomly assigned those respondents to one of two groups. One was an intervention group, where canvassers would engage in discussions like the one Virginia had with Gustavo involving a question around support of trans-inclusive legislation. The others were assigned to the control group, where canvassers would talk to respondents about recycling. All of the participants were then sent surveys at three days, three weeks, six weeks, and three months after their meeting with a canvasser.

But Broockman and Kalla had also added a twist to the experiment. In the survey sent after six weeks, the researchers attempted to destroy any goodwill the canvassers might have sown for trans rights with an anti-trans attack ad. (Ads like these highlight the claim that transgender rights laws will allow predatory males to enter women's bathrooms.)

The results of the six-week survey suggested that the immediate effect of the ad was that it diminished support for a trans-supportive law in both groups. But then at three months, that effect of the ad disappeared for the intervention group — as if they had been vaccinated against prejudice.

In the paper, Broockman and Kalla describe the lasting opinion change between the experimental and control group as comparable to the public's changing attitude toward gay and lesbian people that occurred between 1998 and 2012. "That two decades of opinion change took place during a 10-minute conversation, and it persisted for at least three months — that's a big effect," Kalla tells me.

These results have scientists excited

A few political scientists not affiliated with the project agreed: This work is groundbreaking.

"I think this paper is monumentally important," Betsy Levy Paluck, a psychology professor at Princeton, tells me.

What's most impressive is the design, she says, since most research on political opinions takes place in university labs, not in the real world.

According to Levy Paluck, just 11 percent of all the political science literature on prejudice reduction tests theories out in the real world. "And a lot of those are done in schools — with kids," she says. "We're talking about minuscule proportions that are looking at adults. This [current study] is gold-plated, transparent, big sample —something that we can put some trust in."

Patrick Egan, who studies political attitudes and LGBTQ politics at New York University, is also "tremendously impressed." That an experimenter could produce an opinion change that lasts three months, he says, is "something that has not yet been shown — as far as I’m aware — in psychology or sociology or political science."

I know what you may be thinking: Isn't this all a little too good to be true, again?

Broockman and Kalla are aware their work will be scrutinized. And they want others to check the data, so they've made it and the computer code freely available online. "We fully expect people to start downloading our data and hopefully confirm our results," Kalla says.

And Broockman notes these results are just a "prologue" to a new field of research on prejudice reduction — and not the final say. He hopes other experimenters will attempt to replicate and expand on the study in the coming years.

Deep canvassing is very difficult and very promising

These results come at a crucial time for the transgender rights movement, which is growing in visibility but still facing setbacks in communities across the United States.

Just weeks ago, North Carolina passed a law which, among many things, bans transgender people from using bathrooms that don't match the gender on their birth certificates. Elsewhere in the country, transgender rights are threatened too. As Vox's German Lopez reports, Tennessee may soon pass a transgender bathroom-ban measure.

Activists — and not just those in the LGBTQ community — will likely want to know how to leverage the insight from this study for their own fights. But they should take heed: The "deep canvassing" method Broockman and Kalla tested may now be scientifically supported, but it doesn't mean it's easy or necessarily scalable.

"It's definitely incredibly difficult," says Justin Klecha of SAVE, the LGBT advocacy group in Miami that provided the canvassers for Broockman and Kalla's study. "One is just having the field staff prioritize this work above all else. … You need to build up leaders, and training, and support, which takes time."

Advocates need months to mobilize enough canvassers to meaningfully change votes on a ballot measure. And even then, the intervention may not work on everyone.

But there is more good news to glean from the new Science paper. LaCour and Green's original paper purported to find that only gay and lesbian canvassers could increase support for same-sex marriage. Broockman and Kalla have found that it doesn't matter if the canvassers are transgender or not. "Whether you're trans or not trans, gay or not gay, you're still able to develop empathy with a voter and help them connect the prejudice they've felt in their lives with the prejudice transgender people face," Kalla says.

What's still in question is whether the canvassing technique could be adapted to political topics less rooted in prejudice. "I'm a gay man; for me there were a lot of ways this project felt meaningful for me," Broockman says. But he's also interested in testing it out on climate change, focusing on getting people to think through the problem as it relates to their own lives. Maybe "get people to think about times when they protected their kids," he says, as an example of how to break through to them.

The case for optimism in social science has been strained lately — from outright frauds like the LaCour paper to the ongoing crisis in which psychology papers do not replicate.

But this paper is a certain win: At least in one small, scientifically verifiable way, humans can learn to better understand one another.