Transgender people face up to 25 times greater risk of abuse, assault and suicide compared with other Americans. And some political groups continue to push to repeal existing local protections or enact further discrimination against transgender persons into law. Now new research published in Science today has found that a simple 10-minute directed conversation can have a significant and lasting effect on reducing that prejudice.

The study was conducted in Miami neighborhoods that had supported an earlier anti-LGBT ballot initiative, and it used a process called “deep canvassing” as a way to interact with people to change attitudes. The technique had been developed by the Los Angeles LGBT Center following the defeat of the campaign against Proposition 8, a ballot initiative that made same sex marriage illegal in California. A prominent paper by University of California, Los Angeles, political science graduate student Michael LaCour and Columbia University political scientist Donald Green, published in late 2014 in Science, had found the technique to be unusually successful in changing attitudes toward marriage equality for gays.

David Broockman and Joshua Kalla, then both political science graduate students at the University of California, Berkeley, decided in early 2015 to assess the effectiveness of deep canvassing in changing prejudicial attitudes toward transgender persons in south Florida. But in preparing for that study, “we learned in the process there were a lot of things [in the LaCour paper] that didn’t add up,” says Broockman, now an assistant professor of political economy at Stanford University. That is his polite way of summarizing that LaCour had fabricated much of his data; Green retracted the paper when he learned of the fraud.

Yet Broockman and Kalla believed the deep-canvassing technique had promise, so they applied it with voters in Miami–Dade County, Fla., engaging them in a series of baseline and follow-up attitudinal surveys over the course of months, for which the voters were modestly compensated for their time. The questions covered a wide range of subjects and voters were not told the principal purpose of the research. At one point volunteer canvassers who were not identified as affiliated with the surveys went door-to-door for a conversation with participants; some talked about transgender issues, the “placebo” group talked about recycling.

Traditional canvassing typically engages voters to identify support or opposition for a candidate or issue through a brief series of questions, then quickly moves on. But with deep canvassing, “the canvasser does very little talking—just a like a therapist, only with a voter,” Broockman says. He calls it “a kind of theoretical cousin to cognitive behavioral therapy. It is about the canvasser’s skill in getting the voter to do mental work” in dredging up incidents in their own lives where they were discriminated against. Then the canvasser helps the voter relate their own experience to discrimination experienced by transgender people. The entire interaction took 10 to 15 minutes, and follow-up online surveys measured attitudes at various points for the next three months. “Those conversations led people to be durably less prejudice against transgender people for a least the three months that we have looked at them so far,” Broockman says. “It also increased the core support for including transgender people in nondiscrimination laws.” The deep-canvassing intervention changed attitudes across sociodemographic categories, affecting young and old, Democrats and Republicans, and people of different races. The overall average change in attitude after the 10-minute conversation was roughly equivalent to the degree that social attitudes toward gay people changed between 1998 and 2012.

To see whether the shift in attitude would endure despite hearing opposing messages (as would be common during a political battle), during the online survey six weeks after the canvassing conversation the researchers showed voters one of the transphobic television ads used to overturn legal protections for transgender people in Houston. Immediately after seeing the ad, the voters’ average attitude shifted away from supporting transgender rights—but that effect faded. By the end of the three-month study period, voters’ attitudes had returned to where they were before they saw the attack ad—that is, those who had experienced the canvassing conversation were significantly more in favor of transgender rights.

The results “stand alone as a rigorous test of this type of prejudice-reduction intervention,” wrote Princeton University psychologist Elizabeth Levy Paluck in an accompanying commentary. Previous research generally has focused on measuring prejudice in the field while interventions have been confined to a laboratory setting.

“It is a strong paper that combined a very clever design wrinkle of a field experiment and a survey experiment to show that the effects of the conversation persists over time,” says Green, co-author of the earlier withdrawn paper. Additionally, even when participants were shown a commercial that was hostile to transgender rights, the effects of that ad were transitory, “the opinion did not snap back to the initial view but in fact persisted.”

Mass communication and social media dominate daily life so much that “face-to-face encounters have become increasingly rare—and because they are rare, they may be more memorable” and impactful, suggests Kenneth Sherrill, professor emeritus of political science at Hunter College in New York City and sometimes referred to as the dean of gay political scientists. “If you talk to people face-to-face in their home, you change their opinion more than you do in other ways,” he adds.

“Now that they have demonstrated that this methodology is feasible and deep canvassing seems to be effective, the next step will be to assess the conditions under which it is more or less effective, which kinds of messengers are most effective and for which kinds of subjects are they most effective,” Green says. He wonders how readily political consultants will try adopt the methodology; they may hesitate because of concerns about the feasibility of training and supervising people to carry out deep-canvassing activities. He also notes, with a grain of cynicism, that powerful political consultants generally do not grow wealthy promoting door-to-door canvassing; they become wealthy by taking a commission on expensive mass media campaigns.

The combination of being able to identify people online for a baseline survey, then reach them going door-to-door, and finally resurvey those same people online at a later point has dramatically reduced the canvassers’ need to knock on doors by one or two orders of magnitude, and makes comparison groups much more feasible. “A few years ago it would have cost something like $2 million to do this type of study, so it either wasn’t done or those who did it did very imprecise studies,” Broockman says. “Now with further improvements we have [in methodology], we can do it for more like $25,000.” He believes this cost reduction will lead political scientists and marketers to adopt a deep-canvassing approach to studying attitudes.