Last year, a widely praised study on changing conservative minds on same-sex marriage grabbed headlines for all the wrong reasons. The study’s lead author was discovered to have falsified the results, and the journal Science was forced to issue a retraction.

Now, a new study attempting to truly assess the same type of canvassing work has found even more promising results — this time in reducing bias against transgender people.

The study, published Thursday in Science, found that a single, in-depth, 10-minute conversation by a trained canvasser at the doorstep of a voter resulted in a “substantial” reduction in that voter’s anti-transgender prejudice. It also increased the voter’s support for anti-discrimination laws that protect transgender people.

What’s more, the effect lasted three months — a stunning finding considering that most efforts to change people’s deeply held beliefs have found little to no long-term impact.

“Most attempts by campaigns to influence voters don’t have an impact at all, and the ones that do, the benefit decays in three to five days,” said David Fleischer, director of the Leadership LAB at the Los Angeles LGBT Center, who pioneered the canvassing technique.

The findings could be a ray of hope for LGBT activists who nearly a year after winning the same-sex marriage battle before the Supreme Court are struggling against a wave of state-level laws across the South that they fear essentially legalize discrimination against their community, particularly transgender people.

North Carolina recently passed a law requiring transgender students and others to use the bathroom that corresponds with the sex listed on their birth certificate, and similar legislation is under consideration in some other states. Activists believe that bias against transgender people is behind these bills, and this canvassing technique could be one way to counter that bias.

The findings also represent a sort of personal vindication for Fleischer, a longtime gay rights activist and political consultant. Through his work through the LGBT Center, he had been sending activists door-to-door in conservative neighborhoods in Los Angeles ever since the passage in 2008 of Proposition 8, California’s voter-approved ban on same-sex marriage. (The Supreme Court overturned the ban in 2013.)

The purpose of the canvassing was not to repeat talking points or hand out literature, but to engage voters in deep, personal conversations. The aim was to personalize the issue of same-sex marriage and, he hoped, change voters’ minds on the subject. Certain he was onto something, Fleischer enlisted a young researcher to test the effectiveness of his canvassing method.

The results of that study showed something unprecedented: That these doorstep conversations were changing people’s minds long-term. But the results, it turned out, were bogus. Though canvassers were indeed going door-to-door, it appeared that the researcher, Michael LaCour, had falsified the data.

Hungry to know for sure whether his technique was effective, Fleischer tried again — this time, employing the research team that had debunked LaCour’s study to look at a project underway in Miami aimed at increasing support for an anti-discrimination measure there that included protections for transgender people.

For the new study, the researchers looked at the work of 56 trained canvassers who engaged 501 voters in in-depth conversations about transgender people. Some of the canvassers were transgender while others were not. The conversations lasted about 10 minutes each and were much more intimate than most political door-to-door work; the canvassers showed the voters videos of anti- and pro-transgender ads and probed them about times when they themselves were judged for being different.

The researchers returned to the voters weeks or months later, and using a variety of measures found that voters’ bias against transgender people had diminished significantly, with decreases “greater than Americans’ average decrease in homophobia from 1998 to 2012.”

Fleischer says that the results in fact surpassed those of the discredited study. The original study had concluded that a long-term change in views of same-sex marriage only occurred when the canvasser was gay. The new study found that it didn’t matter whether the canvasser was transgender or not. The effect still lasted for three months.

The study should not be viewed as providing an easy solution to the fights raging in North Carolina and elsewhere over transgender people and anti-discrimination measures, said David Broockman, assistant professor of political economy at Stanford University and the study’s lead researcher. The canvassers employed very specific strategies in their conversations with voters, learned as part of an in-depth training orchestrated by the LGBT Center and SAVE, a Florida-based gay rights group.

“The message is not that there’s a silver bullet,” he said. “This is [a canvassing technique] that took the Leadership LAB a long time to develop, and I don’t think anyone should be under the illusion that they can just go out this afternoon and employ this technique.”

Still, he said, it suggests that voters can be persuaded to change their ingrained beliefs, and that this canvassing technique may be useful in changing minds in a variety of areas, from climate change to racial stereotypes.