clock menu more-arrow no yes mobile

Filed under:

How to talk someone out of bigotry

These scientists keep proving that reducing prejudice is possible. It’s just not easy.

A woman holds a sign that reads, “Spread love, not hate,” while another holds a sign over her head that reads, “Love will always trump hate.”
Hundreds of protesters gathered in New York City after the election of President Trump on November 11, 2016.
Spencer Platt/Getty Images
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.

What does it take to divert someone away from prejudice and toward greater acceptance of others in order to build support for progressive causes? “Deep canvassing,” a relatively new technique, is showing promise — and is backed by rigorous testing from researchers and activists in the field.

One such activist is Vivian Topping, who decided, along with other LGBTQ activists and allies, to try deep canvassing in Massachusetts in 2018, when transgender rights were on the ballot.

Massachusetts voters could choose to keep or throw out a law that banned discrimination based on gender identity. Topping, who’s nonbinary, and others, went door to door. If they met a voter who wanted to get rid of the law, they wouldn’t call them out for prejudice. Instead, they did something more radical: They listened, nonjudgmentally, and began a conversation.

It’s not easy to confront people whose votes would seek to hurt you, and then try to change their minds. “I came out two years ago now, and one of the hardest things for me has been talking with folks who don’t understand [gender identity], and not immediately writing someone off because they don’t immediately get it,” Topping says.

Topping calls this “giving them grace.” It’s a powerful idea: “Giving grace ... means being able to hear someone say something that can be hurtful, and trying to think about how to have a real conversation and connect with them.”

Massachusetts voters chose to protect trans rights, and Topping believes deep canvassing helped. “This tactic is the only thing that has been proven to work on nondiscrimination, so without it we wouldn’t have been able to win,” they say.

Families with transgender children celebrate the new public accommodations bill at the Massachusetts State House on July 11, 2016.
Suzanne Kreiter/The Boston Globe via Getty Images

Giving grace. Listening to a political opponent’s concerns. Finding common humanity. In 2020, these seem like radical propositions. But when it comes to changing minds, they work.

New research tells us changing minds with deep canvassing is not impossible; it’s just very hard. The payoffs are small and incremental, but they are real.

A 2016 study in Science proved it was possible. And now, a new peer-reviewed study — a series of three placebo-controlled field experiments soon to be published in American Political Science Review — replicates the findings and gives us new insights into the conditions for lasting opinion change and reductions in prejudice.

The new research shows that if you want to change someone’s mind, you need to have patience with them, ask them to reflect on their life, and listen. It’s not about calling people out or labeling them fill-in-the-blank-phobic. Which makes it feel like a big departure from a lot of the current political dialogue.

“I think in today’s world, many communities have a call-out culture,” says David Broockman, a UC Berkeley political scientist who has run these experiments with Josh Kalla, a political scientist at Yale University. “Twitter is obviously full of the notion that what we should do is condemn those who disagree with us. What we can now say experimentally, the key to the success of these conversations is doing the exact opposite of that.”

Deep canvassing, explained

Over the past few years, deep canvassing has been adopted by some progressive activist groups looking to not only change minds when it comes to policies on immigration and LGBTQ rights, but also to reduce prejudice toward these groups.

In 2016, Broockman and Kalla showed that a 10-minute “deep canvass” conversation could reduce transgender prejudice for at least three months (you might recall this study was a redo of a previous experiment, from a separate team of researchers, which was retracted due to falsified data).

Topping and dozens of other canvassers were a part of that 2016 effort. It was an important study: Not only has social science found very few strategies that work, in experiments, to change minds on issues of prejudice, but even fewer tests of those strategies have occurred in the real world.

Typically, the conversations begin with the canvasser asking the voter for their opinion on a topic, like abortion access, immigration, or LGBTQ rights. Canvassers (who may or may not be members of the impacted community) listen nonjudgmentally. They don’t say if they are pleased or hurt by the response. They are supposed “to appear genuinely interested in hearing the subject ruminate on the question,” as Broockman and Kalla’s latest study instructions read.

The canvassers then ask if the voters know anyone in the affected community, and ask if they relate to the person’s story. If they don’t, and even if they do, they’re asked a question like, “When was a time someone showed you compassion when you really needed it?” to get them to reflect on their experience when they might have felt something similar to the people in the marginalized community.

The canvassers also share their own stories: about being an immigrant, about being a member of the LGBTQ community, or about just knowing people who are. (You can read the full deep canvassing script here on page 47.)

It’s a type of conversation that’s closer to what a psychotherapist might have with a patient than a typical political argument. (One clinical therapist I showed it to said it sounded a bit like “motivational interviewing,” a technique used to help clients work through ambivalent feelings.) It’s not about listing facts or calling people out on their prejudicial views. It’s about sharing and listening, all the while nudging people to be analytical and think about their shared humanity with marginalized groups.

It’s also quite a departure from standard political canvassing. Typically, in a political canvass, an activist might list a bunch of facts or statistics about why the voter should support their cause. Not so with deep canvassing.

Instead of pelting voters with facts, “we ask open-ended questions and then we listen,” Dave Fleischer, the LGBTQ rights organizer who developed the technique, told me in 2016. “And then we continue to ask open-ended questions based on what they just told us.” 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,” Fleischer said. It is stories, not facts, that are most compelling to people when they’re changing their minds.

Here’s a 2015 video example of deep canvassing. It’s of a real voter and a canvasser from the Leadership Lab, a program of the Los Angeles LGBT Center, which spearheaded this canvassing method after losing the 2008 Proposition 8 ballot initiative in California. The woman in the video starts off ambivalent on transgender issues. But through deep canvassing, the activist is able to turn her around.

Specifically, the canvasser asks the voter to recall a time when he or she was discriminated against. Toward the end of the conversation, the canvasser nudges the voter into thinking about how that experience can relate to the plight of transgender people. The idea is that people learn lessons more durably when they come to the conclusions on their own.

In the video above, notice how the voter starts to come around on the issue when the canvasser asks if she’s ever been on the receiving end of discrimination. She talks about being picked on at work and feeling different. He responds by telling his story of being discriminated against for being gay. It’s a real heart-to-heart between strangers.

And in that moment, he points out that a transgender nondiscrimination law would help people who feel discriminated against at school or work.

”Oh, okay, that makes a lot of sense,” she says.

The video ends like this. “I would totally vote in favor,” the woman says of a transgender protection law. “It’s only right. Let a person be who they are.”

Testing deep canvassing in the real world

In the new study, Kalla and Broockman put deep canvassing through a more rigorous test. Namely: It’s larger, and it targets more issues, both trans rights and policies protective of undocumented immigrants.

The new research also tries to identify the secret ingredient that makes deep canvassing work, and whether versions of it that occur over the phone or through video prompts can be useful as well. (These methods may make it easier to scale up in a bigger campaign.)

The first of the three experiments was pretty much a replication of the 2016 study, but on the topic of rights for undocumented immigrants.

In it, canvassers in three areas — central Tennessee; Fresno, California; and Orange County, California — went door to door and interacted with 2,374 voters in these communities during the runup to the 2018 midterm elections.

“All three places are experiencing demographic change, with a growing and diversifying population of immigrant residents,” says Kim Serrano, the messaging research project manager at the California Immigrant Policy Center. “Tennessee and the Central Valley have been the sites of large-scale workplace raids by ICE [Immigration and Customs Enforcement] in recent years,” she says, “and various cities in Orange County have attempted to ‘opt out’ of the California Values Act.” That’s a state law that limits the collaboration between local law enforcement and federal immigration enforcement.

The experiment, like all the ones in the study, was run a bit like a drug trial: The voters were randomly assigned (before the canvassers even knocked on their doors) to receive either the full deep canvassing conversation treatment, a watered-down version where the voters and canvassers don’t exchange personal stories, or a “placebo condition,” where voters were engaged in a conversation that had nothing to do with immigration. The voters were followed up with by survey one week, one month, and several months after being contacted by the canvassers.

After the canvassing, 29 percent of the people in the placebo condition said they strongly supported policies inclusive of undocumented immigrants. In the full-conversation condition, 33 percent were in support. The effect was durable, too: Three to six months after the conversation, voters who shared their feelings with canvassers in this manner also reported less prejudice toward undocumented immigrants.

The watered-down intervention without the two-way exchange didn’t move anyone to support undocumented immigrants. That’s a new finding.

“Now we can show experimentally that when you take away the two-way nature of the conversation, the effects go away,” Broockman says. It’s this “nonjudgmental exchanging of narratives” that Broockman and Kalla think is the key ingredient in how deep canvassing works.

Keep in mind the media environment the canvassers were working in. Immigration — particularly that of asylum seekers — loomed over the 2018 elections. In the runup, conservative news outlets were blaring headlines about a scary immigrant “caravan” marching north through Mexico to the US southern border. President Trump called it “an invasion,” apparently hoping that by raising xenophobic, dehumanizing fears about nonwhite immigrants, as he had in 2016, he’d help his party win seats in Congress.

In this graph, Broockman and Kalla break down how the canvassing moved the needle on particular questions: whether the government should provide attorneys for undocumented immigrants in legal proceedings; whether the US should grant legal status to people who were brought to the US illegally as children; whether they support deporting all undocumented immigrants; and whether undocumented immigrants should live in fear of daily deportation.

Stanford University

Broockman points out that this graph shows the impacts of deep canvassing among all people who came to the door to answer the canvasser’s questions. It includes those who immediately shut the door in the canvasser’s face. “The numbers get a bit bigger when you just focus on people who actually entered into the conversation,” he says. Among those who started the conversation, there was a 7 percentage point increase for granting legal status to people brought to the US as children, he says, for example.

“This is not just a story of pushing on an open door and taking people who are already Democrats and they just needed a small push,” Broockman adds. “Even as Trump was talking about the caravan, we see that Republicans in our study are moving.”

And like the 2016 study, Broockman and Kalla found it didn’t matter who the canvassers were: They could be members of the impacted communities, or just allies. Both types of canvassers could instigate change.

Why and when does deep canvassing work?

The two other experiments in the study targeted transphobia. In these, researchers included conditions to see whether the conversations could work if conducted over the phone (they did, but it was slightly less effective). In another condition, the canvassers didn’t share their own story, but instead played a video of someone experiencing prejudice and then based the conversation around that. That also worked.

It’s worth noting that some of the results were less strong than those Broockman and Kalla reported in their 2016 paper.

The impacts these conversations had on feelings of prejudice, Broockman admits, are about a third as strong. “When working with new groups, new staff, on a new issue and at bigger scale, I think it’s natural to expect smaller effects,” he says. (It’s hard, he adds, to directly compare the two papers, though, since the 2016 effort focused a bit more on combating prejudice, and this one more so on policy.)

Emile Bruneau, a neuroscientist who studies intergroup conflict at the University of Pennsylvania and was not involved in the canvassing experiments, tells me in an email it is “so promising to see an intervention, any intervention, that has a lasting effect on big social issues.”

What’s missing here, she says, is a theoretical understanding for why the change is occurring. “Without that theoretical understanding, it’s difficult to generalize and use the approach in other settings,” Bruneau says.

It does seem as though the two-way nature of the conversations is essential for the canvassing technique to work. But why? Broockman and Kalla aren’t completely sure. Their main hypothesis is that it works because it’s not threatening. People are resistant to changing their mind during an argument, the hypothesis goes, because it threatens their self-image. Sharing narratives gets around that: The persuasion happens because in talking about themselves, the voters realize a more tolerant attitude is consistent with their self-image.

Broockman says they didn’t set out to find the exact mechanism. “That is just not what we are trying to do here,” he says. Social science experiments are usually conducted on college campuses, in a lab, in contrived scenarios. There’s plenty of work that offers some possible mechanisms by which opinions change. But this work isn’t about that. “One way you could think about our study is as an effort to try to ... use the insights of lab studies in real-world settings,” he says.

(Also worth noting: Deep canvassing has only been tested with progressive causes. Could it be used to wage conservative culture wars? Possibly. Or for issues like the acceptance of genetically modified foods? That’s not known.)

There’s also the question: Is it worth the effort?

The truth is, there’s not much out there in scientific literature on what can change a voter’s mind.

In 2018, Kalla and Broockman published a meta-analysis of 49 experiments that were designed to test whether voters are persuadable by conventional means: phone calls, television ads, traditional canvassing, and so on. In aggregate, it turns out these tactics don’t work at all.

The effects of most efforts to change people’s minds on an issue, if successful at all, tend to fade over time. The impact of television ads, in particular, can fade in just a week. Deep canvassing, it appears from the research, has an effect that can last for several months.

“These deep conversations, I suspect, may be more cost-effective in the long run because the impacts are durable,” Serrano says.

And while the effects may be small, only moving opinion a handful of percentage points among those canvassed may be worth it, too. “I’m a campaign person; you’d do anything for 3.5 points,” says Fran Hutchins, the deputy director of the Equality Federation who worked on deep canvassing efforts reported in the new study. “Think of any of our recent elections — nobody is winning these things by 10 or 20 points. It always comes down to just a few points.”

Do we need more of these conversations in our lives?

There’s a smaller finding nestled in Broockman and Kalla’s new paper, one that might not make headlines but is worth thinking about.

In the experiment on immigration, Broockman and Kalla found that 78 percent of all the people who came to the door when the canvasser rang ended up staying for the entire conversation. And 75 percent of the people who start the conversations with the canvassers share a story about their own lives.

“Those basic numbers tell you something about just how willing most Americans are to have an open conversation with a stranger about these ostensibly divisive issues,” Broockman says.

It’s a reminder that our political opponents aren’t always as rigid or ideologically severe as they appear in our minds. In his work, Bruneau finds that political partisans have a skewed view of how they think their opponents think of them. Which is to say: Republicans assume Democrats dislike them more than they actually do, and vice versa. And it’s this meta-perception, Bruneau finds, that then fuels ongoing conflict and dehumanization.

The activists and scientists I spoke to for this story all agree that you can’t change everyone’s minds.

Topping says, in their experience, deep canvassing works best on people who might be concerned about an issue like transgender people in bathrooms but have never really talked through their feelings. That’s likely a lot of people.

In the age of Trump, there’s a compelling push to call things what they are. When we see racist behavior, we should call it racist and not be euphemistic by calling it “racially charged.” Arguably, there’s a time and place for calling people out, particularly when it comes to powerful, influential people. But maybe not when it comes to our neighbors.

Broockman says this research can at least lend ordinary people a new script when dealing with people in their lives who hold prejudicial opinions. That’s refreshing and useful. These conversations aren’t arguments. In a way, they may be a form of public therapy — for all sides involved.

“This kind of conversation helps me talk to family members who aren’t totally there yet” on accepting their identity, Topping says. “It has taught me patience, and taught me to see people from the most positive view that I can.”