Prejudices are often deep, obstinate beliefs. You've probably noticed this if you've ever tried to change someone's political opinion at a dinner party. But David Fleischer, the director of the Leadership LAB of the Los Angeles LGBT Center, thinks he's found a way to begin changing people's prejudices with just a short conversation.

He and several collaborators struggled for years to get to this point. "We brainstormed every idea and tried every idea, overwhelmingly those ideas failed," he says. And once he thought they had discovered a powerful way to fight prejudice, an enormous scientific fraud perpetrated by other researchers tumbled their progress back a year.

He and his colleagues started the effort in 2009, shortly after the Prop 8 constitutional amendment and struck down same-sex marriage in California. "The LGBT community and our allies were shocked and upset," Fleischersays. "Out of that outrage and despair, people wanted to do something very constructive." He and LGBT Center volunteers began talking to as many people as they could, trying to understand why they lost Prop 8.

Fleischer began getting the sense that just talking and listening to people was making them more accepting of same-sex marriage. "When we were nonjudgmental and vulnerable with them and when we exchanged our lived experiences about marriage and gay people, that's when we started changing people's minds," he says. He calls it "deep canvassing."

The conversations are short, usually 10 to 20 minutes long. The canvassers don't try to build rational arguments for why someone should think one way or another. The goal is to share personal stories about times when the voter and the canvasser felt attacked or discriminated against, says Nancy Williams, a volunteer canvasser with the LGBT Center. "There's something special about caring about why [people] feel the way they do. You can connect to their values in that way."

Williams says around the time she began volunteering with the LGBT Center, she canvassed a man who didn't support nondiscrimination protection for transgender children. "He kept saying high school is hard for everybody. He was thinking as if trans kids don't have all the other things kids have to deal with — on top of being transgender," Williams says. "I told him that my father is fighting cancer right now. There were days it was just hard to face that and everything that I was going through with my transition at the time. I felt like I could deal with facing the possibility of losing my dad and I could deal with the possibility of being rejected by the people I loved. But not both."

Williams says the man said he hadn't considered what that might be like. But he said he's been bullied himself. That, Williams thinks, might have helped persuade him to be more supportive of nondiscrimination measures.

The canvassers thought the conversations were changing people's minds, but Fleischer says he couldn't know if it was working without independent verification. He enlisted a graduate student at UCLA named Michael LaCour to see if there was a measurable effect. LGBT Center volunteers went out to canvass hundreds of people. LaCour and an advisor, political scientist Donald Green at Columbia University, published the findings in Science in 2014. The study got a huge amount of attention.

But things fell apart when David Broockman and Joshua Kalla, then graduate students at the University of California, Berkeley, tried to use LaCour's methods to study prejudice against transgender people. They, along with Yale statistician Peter Aronow, discovered that LaCour's work was almost a complete fabrication. Green, who was unaware of the deception, quickly requested a retraction. LaCour denied the misconduct in a statement to The New York Times last year.

The news crushed Fleischer. "That was a real shock," he says. He called everyone involved with the work and every reporter who had covered the study. "I needed to be sure people knew that we no longer had evidence. Mike LaCour had not assembled [it]. He had lied to us. Taken advantage of us. And I also wanted to point out to people we were not going to give up."

After the dust settled, Broockman and Kalla went on with their experiment on transgender prejudices. LaCour's misconduct only made them more determined to do the study for real. "There were all these volunteers who gave their Saturdays [to do the experiment]," Broockman says. "We had a certain sense of responsibility."

They sent out surveys to thousands of homes in Miami, asking people to answer questions that included how they felt about transgender people and if they would support legal protection against discrimination for transgender people. Then volunteers from SAVE, an LGBT advocacy organization based in Florida, visited half of the 501 people who responded and canvassed them about an unrelated topic, recycling. Volunteers went to the other half and started the conversations that Fleischer thinks can help change minds.

After the canvass, the study participants answered the same questions about transgender people that they had answered before the study, including how positively or negatively they felt towards transgender people on a scale of 0 to 100. Those who had discussed prejudice they'd experienced felt about 10 points more positively toward transgender people, on average.

Broockman says that public opinion about gay people has improved by 8.5 points between 1998 and 2012. "So it's about 15 years of progress that we've experienced in 10 minutes at the door," he says.

Three months after the canvass, Broockman asked participants to fill out the survey again. They still felt more positively about transgender people than those who had gotten the unrelated canvass. "[That's] the moment I backed away from my monitor and said, 'Wow, something's really unique here,' " he says. If the effect persists, Broockman says, the technique could be used to reduce prejudice across society.

That doesn't mean everybody came away feeling more positive about transgender rights. Kalla says some people came away from the canvasser feeling very differently and some people not so much at all. And an uptick in 10 points on a feeling scale of 0 to 100 doesn't sound like an epiphany. There wasn't, however, any indication that those who started out with very negative feelings about transgender people were particularly resistant to the conversation. Broockman and Kalla published the results in Science on Thursday.

It is a landmark study, according to Elizabeth Paluck, a psychologist at Princeton University who was not involved with the work. "They were very transparent about all the statistics," she says. "It was a really ingenious test of the change. If the change was at all fragile, we should have seen people change their minds back [after three months]." There are very few tests of prejudice reduction methods, and Paluck says this suggests the Los Angeles LGBT Center's approach is actually far more effective than previous efforts, like TV ads.

There might be a couple of reasons for that. Broockman, now an assistant professor of political economy at Stanford University, says asking someone questions face-to-face like, "What are the reasons you wouldn't support protections for transgender people, or what does this make you think about?" gets them to begin thinking hard about the issue. "Burning the mental calories to do effortful thinking about it, that leaves a lasting imprint on your attitudes," he says.

Empathy may also be a factor. "Canvassers asked people to talk about a time they were treated differently. Most people have been judged because of gender, race or some other issue. For many voters, they reflect on it and they realize that's a terrible feeling they don't want anyone to have," Broockman says.

The study's conclusions differ from the conclusions of the LaCour's falsified study from 2014 in one crucial way, Broockman says. LaCour claimed that there was only an effect from the deep canvass if it came from someone who was LGBT. "We found non-trans allies had a lasting effect as well," Broockman says. That means canvassing is much more about conversational skill rather than identity.

It will take more studies and replications of this study before scientists know exactly what is influencing people's opinions. But for now, the findings are a relief to David Fleischer. "To go into it with high hopes and then get this really bad piece of news, then to go forward anyway and have the accurate results? What a roller coaster of emotions," he says.

Copyright 2016 NPR. To see more, visit