This article is from the archive of our partner National Journal

Dave Fleischer wants to show me that it's possible to change a voter's mind about a topic as controversial as abortion. Never mind that the scientific paper that proved this theory to be true was redacted in controversy—its underlying dataset appears to have been completely falsified—just a few days ago.

Fleischer presses play on a video. On screen, a man approaches a woman stepping out of a red SUV on the driveway of her Los Angeles county home, in an area that tends to vote down abortion-access measures when they come to the ballot.

He wears a slim-fit T-shirt, has a short beard, and is carrying a clipboard. He looks like the type who might stop you on the sidewalk and ask if you have a minute to spare for the environment. But he's not there for her signature or money. He wants her opinion.

"I don't believe in abortion," she says, flat-out. "Maybe in the case of rape, but otherwise, I don't believe in abortion."

This is where a typical political canvasser might spew talking points—how unplanned babies place an unfair economic burden on mothers, how a woman has a right to decide what's happening to her body. But he doesn't. He asks a personal question instead: Does she have any children? Grandchildren?

Then, something happens. The woman starts to open up. A 20-minute heart-to-heart conversations follows. She reveals one of her sons was in a terrible marriage; his child was unplanned. "I don't know if keeping this child was the right idea," she admits. "My advice—and even though I don't believe in abortion—was to have an abortion. Because I didn't think it is fair to bring that baby into that environment."

"Sounds like you got a conflict," the canvasser says.

"I don't think there is a right answer," she says. "I really don't."

That's a position that's a long way off from "I don't believe in abortion."

Fleischer, 60, has worked in politics for 40 years, and heads outreach efforts at the Los Angeles LGBT center. Growing up in a small town in Ohio, Fleischer didn't come out as a gay man until he was 25. "I've spent my whole life trying to figure out how to reduce prejudice," he tells me.

Fleischer has been refining canvassing methods since Proposition 8 appeared on the ballot in California in 2008. His work is about convincing voters their beliefs—beliefs they might vote on when it comes to a ballot measure—are actually more conflicted when they are prompted by canvassers to think more deeply about them. "When was the last time you changed your mind on anything?" he asks me. "I think you are going to discover you did not change your mind because somebody else—I'm just going to say—bitch slapped you with a statistic."

Fleischer shows me two other similar videos of him and his canvassers approaching California voters. In one, a man who disagrees with transgender issues eventually admits it may be rooted in his upbringing in a bigoted family. In another, a woman reverses her stance on abortion after the canvasser tells her a story about her own abortion.

"If you want a voter to change their mind," Fleischer says, "the most useful thing we do by far is that we elicit their real-lived experience. We ask open-ended questions and then we listen. And then we continue to ask open-ended questions based on what they just told us."

Until recently, Fleischer had the scientific evidence to say his methods worked.

Late last month, the journal Science retracted a 2014 study that found that canvassers who were gay and following Fleischer's method could influence voters to support same-sex marriage more than canvassers who were straight—a conclusion that's now in doubt.

Fleischer was not involved in the data collection himself; Michael J. LaCour, a doctoral student at the University of California (Los Angeles) was studying Fleischer's canvassing methods on his own.

Donald Green, a well-regarded political science professor at Columbia University (people say he wrote "the bible" on field experimentation in political science)—oversaw the project. When it was published in Science, a premier peer-reviewed journal, it was the type of blockbuster study that was covered by all the major newspapers, and recently was featured in a segment of This American Life.

The findings were irresistible. What amazed people about LaCour's results was not only that an interaction with a gay canvasser led to an increase in support in gay marriage in voters, but that the effect persisted for those voters over a nine-month period. Most field experiments that attempt to change voters' minds backfire, often spectacularly. "The Science paper is almost like an existence proof that attitude change is possible," Green told me in April, more than a month before the fraud was revealed.

LaCour faked all the data (he now says he deleted the datasets, so they can't be reviewed) and lied about his funding sources. When Green found out, he immediately asked for a retraction, which Science granted.

"The greatest frustration is is that Mike LaCour chose ... from the beginning he knew he wasn't going to measure our work," Fleischer says. "Which is astonishing to me because we have a beautiful thing to measure."

It's still worth measuring. The fraudulent results are an indictment of the data, not of the hypothesis.

Fleischer's canvassing method is a widely different approach to most political messaging. Increasingly, campaigns don't care about persuading voters, and instead focus on targeting voters who already agree with their platform.

The canvassing study is a real-world test of what's known as the contact hypothesis: The idea that just by being around different people, we become more accepting of them. The classic instance is when a father becomes accepting of gay people if he son comes out of the closet. A racist thinks more fondly of African Americans after working with a black person.

A forthright version of the study also would put to test a question that shadows much research in political science: Is rational political discourse among humans even possible? Do we have the capacity to change our minds?

There are two theories that explain why political persuasion is so difficult. One is called motivated reasoning—that's the theory that says as partisans, we're fighters for our political teams. We actively resist information that might hurt our worldview.

But another theory lends more hope,though it is less supported in the scientific literature. It's called bayesian reasoning, which theorizes that even though we come to a conversation with ideologies, we're still impressionable. This theory posits that when partisans are confronted by inconvenient information, they do process it. It just doesn't move them very far in the other direction. It could take a lifetime of evidence to slowly nudge a climate change denier into believing in it. But according to bayesian reasoning, they can get there—eventually.

Alexander Coppock, a doctoral student at Columbia who often works with Green, has recently put this idea to trial in an online experiment that testedthree controversial topics: the death penalty, minimum wage, and climate change (it's currently under peer review). He found that regardless of the ideological position people began with, when presented with evidence for one side or another, people's feelings swayed slightly in that direction.

"Suppose we're measuring the support for punishment on a scale of one to seven, where one is 'I really oppose it' and seven is 'I really love capital punishment," Coppock says. "It's not the case where you take people who are sevens and turn them into people who are ones. We're taking people who are fives and we're moving them to 4.5."

The conclusion from Coppock's study still has a large limitation: It didn't take place in the real world. That's why studies like the Science article are needed, to see whether the theory can be proven when it comes to human-to-human conversation.

David Broockman, the 26-year-old graduate student who helped uncover LaCour's fraud, is testing Fleischer's canvassing methods on the issue of transgender rights in Miami.

Broockman knows that his replication will probably not be the showstopper the LaCour study was. Still, the thought of succeeding at it is tantalizing. With further study, it could be that Fleischer's methods are not just a useful tool for persuasion on social and reproductive matters. It could be his methods are a generalized tool to combat partisanship. "We want to know if it works for immigration, for gun policy, all of these issues that apparently divide people on the political spectrum," Coppock says.

If Fleischer's methods work, that means we have at least one scientifically valid tool to increase dialog between opposing sides of an argument. It sounds preposterous. But it's an idea worth testing.

"I've had the same experience you had watching those videos," Broockman says. "Some of these conversations are really powerful. So I really don't know [what the results will be]. There really isn't anything out there like what Dave's group does. I'm actually open to the idea that it might even be more effective than the LaCour and Green study. That's the weird irony. It might have understated the effect. We don't know."

This article is from the archive of our partner National Journal.