It Turns Out a Brief Conversation Really Can Change Minds on LGBT Issues

Photo: Whitney Keller/The Herald-Sun via AP

Recently, there has been a spate of state-level attempts to make life harder for transgender people. At the moment, about 7 states are actively debating laws that seek to bar transgender people from using the public bathroom that lines up with the gender they identify with, often under the guise of “protecting” women and children from sexual predators who want to sneak into women’s bathrooms.

Some states have already passed, or come close to passing, these sorts of laws. In North Dakota, for example, a bill targeting transgender kids was vetoed by the governor last month. In North Carolina, on the other hand, the bathroom panic won out: The state passed, and the governor signed, a bill forcing residents to use bathrooms that line up with the gender listed on their birth certificate, and preventing towns from passing nondiscrimination ordinances on this issue. South Carolina is now considering a similar law.

While the passage of these sorts of laws has been met with a backlash — New York and some other states have cut official travel to North Carolina, and the NBA is hinting that it might move next year’s All-Star Game, which is slated to be played in Charlotte — it’s hard to deny that the bathroom fears are, in many cases, proving effective. In much the same way that earlier anti-gay-rights activism successfully riled up straight people with fears of gay teachers converting their kids or teaching kindergartners about anal sex, the bathroom campaigns are designed to hit people in a vulnerable, visceral place.

How do you convince people not to fall for this sort of fearmongering? There is, after all, no evidence of some evil plot to use pro-transgender laws to help sexual predators have easy access to victims. From a social-scientific standpoint, it’s an interesting question — researchers haven’t really found consistent ways to get people to become less fearful and more tolerant of so-called “outgroups,” especially ones they don’t interact with on a frequent basis.

A new study in Science, though, provides evidence for a promising approach. It’s a study with a fascinating, tangled backstory and, potentially, massively important ramifications not only for the fight over transgender equality, but for civil rights in general.


When news of the Michael LaCour scandal broke last spring, it was easy to see it as the end of a once-promising line of inquiry. That scandal, you may recall, involved the revelation that LaCour, then a graduate student at UCLA, had fabricated data for a blockbuster study he co-authored with Don Green, an esteemed political scientist at Columbia, which was published in Science in December of 2014.

LaCour and Green’s paper purported to find that a brief conversation with a canvasser had a profound impact on Californians’ views about gay marriage — but only if that canvasser was gay (and shared that fact with the resident they were speaking with). It was such a striking result, and one that flew in the face of so much prior research into the difficulty of persuading people to change their minds about hot-button political issues, that the study made nationwide headlines. But in a story first told in its entirety by Science of Us, a team of two UC-Berkeley graduate students, David Broockman and Josh Kalla — with some last-minute help from the Yale political scientist Peter Aronow — gradually came to realize that LaCour and Green’s results were, in fact, too good to be true.

LaCour, it appears, simply made up the data by adding noise to a preexisting data set, and even fabricated an email from a nonexistent employee of the survey company he had ostensibly hired for his research. After Broockman, Kalla, and Aronow posted their findings online, the paper was retracted; other instances of LaCour’s dishonesty were uncovered; and LaCour’s job offer to be a professor at Princeton was rescinded.

LaCour might be out of academia, but this line of research, it turns out, lives on. Part of the reason Broockman and Kalla were able to discover the irregularities that toppled the paper was because they were looking to run a canvassing study of their own, in Miami. Their resulting paper has just been published in Science, and it shows that canvassing has some serious potential to nudge people toward supporting transgender rights — though there are some important differences between their findings and what LaCour claimed to have discovered.

For the study, Broockman, who is now an assistant professor at the Stanford Graduate School of Business, and Kalla looked at a canvassing effort launched by the Los Angeles LGBT Center and SAVE, a Floridian LGBT organization, last year. Following the passage of an antidiscrimination ordinance “protecting transgender people from discrimination in housing, employment, and public accommodations” in December of 2014, the organizations, wary of a potential backlash, sent volunteer canvassers in Miami-Dade County to knock on doors and talk to residents about transgender rights.

The encounters with residents were structured in a very specific way. Canvassers were told, after introducing themselves, to inform residents that they might soon have to vote on whether to protect transgender people from discrimination (since there might be a referendum to repeal the ordinance), and to ask them what their views were on the law that had just been passed. The canvassers were trained to express interest in the residents’ opinions, but not to show signs of judging them for what they were saying one way or another.

After hearing a resident’s opinions on the law, canvassers shared a video in which proponents on both sides expressed their views. The subject was asked to react to the video, sparking a bit more conversation with the canvasser. Then — and this is key — the canvasser asked the subject whether they’d ever experienced stigma or negative judgment like a transgender person might. If necessary, the canvasser gave examples from their own life to help fuel the conversation — some of the canvassers were transgender, and at this point they might reveal this and tell the subject a bit about their experiences being judged. Finally, at the end of the conversation, the subject was asked again for opinions about the law and transgender people.

Here’s a video of one of these canvassing conversations taking place in Los Angeles:

All this applied to half the total participants in the study; the other half was assigned to a placebo group, and its members were visited by canvassers who talked to them about recycling.

To test the effects of these encounters, Broockman and Kalla looked at surveys filled out by the respondents both before they were contacted by canvassers (the surveys included a bunch of other questions as well, shielding their true purpose) and at multiple points after: three days, three weeks, six weeks, and three months. As is true with any survey with multiple waves, at each step there was some drop-off — the total number of respondents surveyed started at 429 and dipped to 385.

Broockman and Kalla found that the encounters worked. Three months after the canvassers visited — the latest the results were tracked — residents who had had a conversation about transgender rights were measurably “warmer” in their feelings toward transgender people as compared to those in the placebo group. Before the canvassing, the two groups felt about equally toward transgender people on a 101-point “feelings thermometer,” but three days after, those in the active condition felt about ten points warmer, on average, “an amount larger than the average increase in positive affect toward gay men and lesbians among Americans between 1998 and 2012 (8.5),” the authors write. Three months later, the difference was still 9.2 points. All from a conversation that lasted, on average, just 10 minutes.

Interestingly, it didn’t matter whether the canvasser was transgender or cis-. This had been the big, headline-grabbing takeaway of the fake LaCour finding: that there was something so profound about talking to a member of the outgroup in question — gay people, in that study — that it had an effect that couldn’t, apparently, be replicated by assigning straight canvassers to talk to voters. In reality — when it comes to transgender issues, at least — that’s not the case. “The fact that a non-transgender canvasser could have these effects shows you that it’s not just about the image of the canvasser or changing the face of the issue,” Broockman told Science of Us.

So what is it about? Broockman and Kalla think two mechanisms are at work here: “active processing” and “analogic perspective-taking.” Active processing simply means getting someone to actually engage with an issue in a thoughtful way — here Broockman referenced Daniel Kahneman’s famous divide between system 1, which is gut-level, quick-response thinking, and system 2, which involves deeper thought and rumination. Active processing, which the researchers think is triggered by asking respondents to explain their opinions on transgender people, nudges people into system 2 thinking. Analogic perspective-taking, meanwhile, is simply the act of putting yourself in someone else’s shoes and considering similarities between you and them.

In the mixing of the two is where the researchers think the magic lies: “Active processing signals that the perspective-taking you’re doing is not a shallow exercise,” said Broockman, “but one where you’re engaging with a lot of cognitive effort.” You’re not just saying, Yeah, yeah, transgender rights are great — got it. You’re in a place where you really can understand, in an intellectual way, that there are some similarities between a transgender person’s experience and your own.


These results are, in a sense, the end of a long and circuitous roller-coaster ride for Dave Fleischer. Fleischer, a 61-year-old Harvard Law School graduate, heads the Los Angeles LGBT Center’s Leadership LAB, which, over the years, has developed and refined the approach used in the Broockman and Kalla study.

LaCour’s fabrication hurt a lot of people, but there’s a case to be made that Fleischer was hurt just about as much as anyone, since the original Science paper, too, had been (ostensibly) based on Leadership LAB canvassing. Fleischer had been quite eager to get the Center’s canvassing methods externally evaluated, and had naturally been thrilled at LaCour and Green’s findings. On the morning of May 19 of last year, he was scheduled to meet with LaCour to discuss the next phase of their collaboration: LaCour was about to submit a paper based on the Leadership LAB’s abortion-rights canvassing work. Instead, he got a text message from LaCour: “Lynn says I have to reschedule.” ‘Lynn’ was Lynn Vavreck, his dissertation advisor at UCLA, who around that time confronted him with the information Broockman, Kalla, and Peter Aronow had brought her.

Part of the reason it had been such a rush for Fleischer to hear how well the Leadership LAB’s technique had performed was because it fit into Fleischer’s larger view that something is amiss, persuasionwise, in the world of LGBT advocacy. In a New York Magazine conference room earlier this week, he expressed palpable frustration with the output of the slick D.C. consultants often employed by LGBT organizations to convince straight people to be more tolerant.

He opened a Mac whose hard drive was stuffed with what he viewed as lowlights of decades of lackluster advocacy (as well as some positive counterexamples). “There are a variety of ways in which we sort of kid ourselves,” he said, as he cued up an ad from California’s Proposition 8 referendum banning gay marriage in that state, which at the time of its passage in 2008 was a demoralizing hit to the national LGBT community (it was subsequently overturned in court). In the 30-second spot, produced by opponents of the referendum, Samuel L. Jackson compared Prop 8 to other past acts of official discrimination against Californians of various stripes, and told viewers proponents of the proposition wanted to “eliminate fundamental rights.”

“Did you notice what they did that made them feel like this was gay?” Fleischer asked me after the ad was over. No, I replied, since Jackson hadn’t even mentioned gay marriage. “Of course you didn’t catch it!” said Fleischer. “Neither did a single fucking voter. We’ll watch it again.” When we did, he pointed out two ever-so-brief still shots of gay couples comprising maybe a second of total screen time (look at second 23 in the YouTube video of the ad). “Gay people! Gay people!” Fleischer enthused, sarcastically.

At the time, proponents of Proposition 8 were warning nervous Californians that it would lead to all sorts of dire consequences, including discrimination against religious people and organizations. To Fleischer, it makes no sense to respond to such a potent attack on gay rights by relegating actual gay people to an eye-blink’s worth of screen time.

Or take a more recent example: the fight over the Houston Equal Rights Ordinance (HERO), which would have banned discrimination on the basis of sexual orientation, gender identity, and other categories in Houston (technically, the vote was to veto an antidiscrimination ordinance that had already been passed by the city council).

In the run-up to the vote last November, the anti-HERO ads were brutal. In one, a woman’s voice laid out dire scenarios about what would happen if HERO passed, as spooky footage of dark, dreary bathrooms rolled: “Any man at any time could enter a woman’s bathroom simply by claiming to be a woman that day. No one is exempt: Even registered sex offenders could follow women or young girls into the bathroom, and if a business tried to stop them, they’d be fined.” The last ten seconds of the ad consist of a young girl entering a bathroom stall, at which point another figure — clearly meant to be a man — exits the stall next to her and accosts her, presumably to assault her.

The problem, as Fleischer sees it, is that when people have very real fears about, for example, men dressing up as women so they can rape women in bathrooms, if you want to beat back this fear, you had better damn well address it directly. If one side is making the emotional argument that the cross-dressing rapists are coming, and the other responds with thin pablum that seems geared at misdirecting or obscuring, the haters are going to win every time.

But all too often, it’s the saggy approaches that win out. “We will talk about anything except what is really on people’s minds,” said Fleischer. The four pro-HERO ads produced during the run-up to the vote in November fit this pattern exactly; they were positively milquetoast by comparison to the scary bathroom spots. One of the pro-HERO ads made the debate out to be really about veterans’ rights, not transgender people’s rights; a second offered a cheery series of platitudes about how discrimination is bad and inaccurately implied that the defeat of HERO would relegalize racial discrimination; a third attempted to rebut the bathroom claims simply by pointing out that it’s “already illegal” to sneak into a bathroom to harass or harm someone; and the final one included a pastor and a reverend (and their wives) speaking in generalities about the bill’s importance.

Fleischer wasn’t the only one unimpressed with this approach. “The anti-HERO opposition used ads which portrayed transgender people as creepy and scary-looking, while implying they were sexual predators,” said Brynn Tannehill, an advocate for transgender rights who writes for the Advocate and the Huffington Post. By contrast, “The messaging by pro-HERO groups tried desperately not to talk about transgender people and public accommodations, which looked like they were avoiding the issue.”

Indeed, none of the ads included a transgender person, and none of them addressed, head-on, the reasons the bathroom fears were ill-founded.
HERO lost by a lopsided vote of 61–39. “The ads that our side aired missed the opportunity to educate Houston voters about who transgender people are,” said Fleischer. “They missed the opportunity to humanize transgender people. And since these ads represented the lion’s share of the more than $3 million that our campaign spent, it means that the campaign itself missed the opportunity to move things forward and reduce prejudice against transgender people.” Now, can it be proven that the ads were the reason the bill faltered? No, and Tannehill, for her part, said she wouldn’t put it in the top few reasons. But to Fleischer, it’s an obvious starting point for an autopsy. “We outspent the other side 6–1, we started out ahead in the polls, and we lost 61–39,” he said. “How much more evidence do you need before you say kaddish and move along?”

So how do ads like this get made over and over again, despite the lack of hard evidence they hold any power to sway swayable voters? Fleischer believes recent political history is partially responsible. To oversimplify a bit: Fleischer explained that when the LGBT movement first started to professionalize around the late 1970s, it turned, as any burgeoning movement would, to the professional class of consultants in Washington, D.C. It was time to make ads to convince straight people to embrace gay rights. The problem, as Fleischer sees it, is that these consultants had — and have — mostly built up successful-seeming careers by making ads for incumbent political candidates, often racking up impressive records in part because incumbents get reelected in the U.S. at staggering rates, not necessarily because anything about the ads themselves was particularly persuasive.

And if you’re making an ad for a candidate you expect to win, the incentives are going to point to a specific strategy: Namely, you’re going to want to be relatively risk-averse, to not offend or annoy anyone. It’s the other guy (or gal) who has to take risks, given the uneven playing field between an incumbent and a challenger. “They’re so unused to working for an underdog,” Fleischer said of many D.C. consultants, and a strategy built for an incumbent will only get you so far in the realm of LGBT rights. When it comes to those fights, after all, straight and cis voters are aren’t choosing between a familiar incumbent and a less-familiar challenger, but rather are voting on bills that, they are told by demagogues, could lead to their children being turned gay by a teacher, or raped by cross-dressers, or whatever else. “The truth is they don’t know a thing about how to win a thing for an underdog, or how to help an underdog cause,” Fleischer said of the consultants. “And the LGBT cause, when we’re on the ballot, is always an underdog cause.”

All of this explains why Fleischer takes the concept of external evaluation so seriously. Many political communications professionals cash check after hefty check, from client after client, without facing any rigorous tests to prove that their approach is working, and he doesn’t want his organization to fall for this sort of self-delusion.

Broockman pointed out that rigorous evaluation is not the norm for people who persuade for a living. “There are many campaigns that are doing some sort of inferior form of research, be it proprietary research where the incentives aren’t aligned” — situations in which Organization A pays Organization B to “evaluate” its work, but everyone involved knows that the desired outcome is a glowing report about how effective A’s approach is — “or polling or focus groups that don’t really tell you what real-world effectiveness is,” explained Broockman. “I think that the credit the LGBT Center should get is not just their persistence in developing this prejudice-reduction message, but the fact that they proactively reached out to academics to be measured.”

Being tested in this manner is a risk, simply because it might show your approach doesn’t work. And in fact, Broockman and Kalla released a separate paper today showing that an abortion-rights message developed by the Leadership LAB didn’t work, according to data drawn from a chunk of canvassing encounters in Los Angeles geared at “seeking to increase support for safe and legal abortion and attempting to reduce stigma towards women who have had abortions.” (In an email, Kalla offered some informed speculation as to why this canvassing didn’t work: “One possible difference between the transgender and abortion studies is that issues around the rights of transgender people are relatively unknown to the American public, while on abortion, voters tend to have a better understanding. Because voters approach abortion with more preexisting considerations, this could make attitude change on abortion more difficult.”)


At first glance, it might seem like apples and oranges to compare TV ads to canvassing. They’re two very different approaches to persuasion, after all. But what matters most here, probably, is the system 1 and system 2 thing. Think about that bathroom commercial — it is decidedly not an attempt to get voters to sit carefully and thoughtfully with an issue and make a decision based on anything like rational analysis. No: It’s a limbic appeal, one designed to get them to recoil in horror at the prospect of their loved one being targeted by a predator.

It may be the case that you can’t fight this sort of fearmongering with 30-second spots. Broockman, for his part, said that if he were running a four-year campaign in the run-up to a vote on an LGBT issue, he’d spend far more money on generating volunteers who would be willing to go door-to-door than on TV ads. That’s partly because he saw firsthand how canvassing disarmed Miami-Dade voters’ bathroom fears, how it shifted them from fearful system 1 reacting to thoughtful system 2 thinking. “At the beginning [of a canvassing encounter] people reacted to [the bathroom claims] very strongly,” he said. “And you bring it up at the end and you ask a simple question — What exactly do you worry about happening? — and quite often, once voters are in a mindset where they’re thinking things through more carefully, often they’re like, Oh, this is ridiculous.” On the other hand, maybe TV ads featuring real-life transgender people talking about how they just want to be able to use the bathroom would trigger something like the same effect — it would certainly be worth a shot, if these groups are going to be dumping millions into TV advertising anyway. But at the moment, Broockman, Kalla, and Fleischer are excited about the prospect of spreading the canvassing gospel.

The next step will be figuring out exactly why it works. As Fleischer and his team developed his approach over the years, they didn’t do so with an eye explicitly on the psychological literature. Rather, their approach was a bit more intuitive and experience-driven, an iterative process of countless canvassing campaigns, of videotaping encounters, and of developing a sense of what was working and what wasn’t — with a healthy dose, it should be said, of regular internal assessment. As Fleischer put it, the Leadership LAB’s m.o. in developing its canvass approach has been “trial and error, a lot of error, but with enough self-awareness and honesty that we kept trying new ideas when old ones clearly were not working.”

So the active-processing/analogic-perspective-taking theory is just that: a theory. It’s an attempt to take what the canvassers were doing and, as Broockman put it, “map it onto the language of psychology.” Future experiments, then, will try to isolate which aspects work and which don’t — and then all of that will be carefully measured, replicated, and so on. “I do think we’re going to have many, many studies like this in the years to come,” he said.

In the meantime, of course, there’s going to be a lot of questionable advocacy going on, leading — arguably – to wasted money, dead ends, and failed electoral battles. But one of the most important things about this paper is that it will help cement the norm that attempts to persuade people on civil-rights issues should be measured rigorously and independently. Gut isn’t enough.

The field of political persuasion, after all, has its own entrenched incumbents — that is, the consultants who bring home hefty paychecks without really proving what they’re doing works — and maybe it’s time they faced a stiff challenge.