Fake Science, Real Consequences
This morning, Social Science Twitter is consumed by the discovery of fraud in a very widely-circulated political science paper published last year in Science magazine. “When contact changes minds: An experiment on transmission of support for gay equality”, by Michael LaCour and Donald Green, reported very strong and persistent changes in people’s opinion about same-sex marriage when voters were canvassed by a gay person. The paper appeared to have a strong experimental design and, importantly, really good follow-up data.
Can a single conversation change minds on divisive social issues, such as same-sex marriage? A randomized placebo-controlled trial assessed whether gay (n = 22) or straight (n = 19) messengers were effective at encouraging voters (n = 972) to support same-sex marriage and whether attitude change persisted and spread to others in voters’ social networks. The results, measured by an unrelated panel survey, show that both gay and straight canvassers produced large effects initially, but only gay canvassers’ effects persisted in 3-week, 6-week, and 9-month follow-ups. We also find strong evidence of within-household transmission of opinion change, but only in the wake of conversations with gay canvassers. Contact with gay canvassers further caused substantial change in the ratings of gay men and lesbians more generally. These large, persistent, and contagious effects were confirmed by a follow-up experiment.
At the time, on the Washington Post’s Monkey Cage blog, Andrew Gelman noted that the magnitude of the effects reported was unusually large:
What stunned me about these results was not just the effect itself—although I agree that it’s interesting in any case—but the size of the observed differences. They’re huge: an immediate effect of 0.4 on a five-point scale and, after nine months, an effect of 0.8. A difference of 0.8 on a five-point scale . . . wow! … And this got me wondering, how could this happen? After all, it’s hard to change people’s opinions, even if you try really hard. And then these canvassers were getting such amazing results, just by telling a personal story?
Gelman went on to hypothesize a plausible story about changes in opinion on this topic specifically. But it now appears, based on the technical report by three social scientists who had started out intending to do a follow-up study, and accompanying retraction letter from the second author, that the follow-up data were faked, apparently by LaCour alone.
When something like this happens it raises many issues internal to academia, from the relative role of the authors involved, to the need for available and replicable data, to the often unrecognized importance of simple honesty in science. As a social scientist I worry most about the quality of the frauds we don’t spot. Science is often bitterly competitive but it depends on honesty. It is not set up to weed out liars. Imagine what research, or talks, or conferences would be like if you had to routinely question not simply the quality or competence but the actual honesty of speakers. The same goes for supervision. Consider having to check not just the quality of your grad students’ work, but whether they were lying to you about their data. Much of what we do would become simply impossible.
If it’s true that the data don’t exist, I expect Michael LaCour will lose the job he just got at Princeton. I would also not be surprised if his Ph.D—if he’s defended it—is revoked by UCLA as well. But this isn’t just a matter of internal incentives and social-scientific practice. That paper got a lot of public attention, and people both heard about it and tried to learn from it. This week sees Ireland enter into the final week of campaigning on a constitutional referendum that, if it passes, will make Ireland the first country in the world with a constitutional right to same-sex marriage enacted by popular vote rather than judicial decision. Things are looking good for the Yes campaign at the moment, which is remarkable considering how historically conservative the country has been. (In 1995, a referendum to legalize divorce passed by less than ten thousand votes out of more than 1.6 million.) As my cousin Ronan Palliser reminded me this morning, and as reported in the Irish Times, the results of the LaCour and Green paper were used as a template by those organizing the Yes campaign:
A follow-up study by UCLA showed that not only did people change their minds, they stayed the course over the long term and even persuaded others to do the same. It was a game-changer with a measurable increase in support among that cohort. The This American Life podcast [which reported LaCour and Green paper] has been widely distributed by the Yes Equality campaign to its supporters and branches throughout Ireland. It has also provided a template for its approach to the campaign for the May 22nd referendum. …
Visibility is one thing. Sure, part of the Yes campaign is mobilising its base which means among other things “get out the vote” and ensure younger supporters register. Beyond that, though, the campaign has very deliberately reached out to another group which would not be classed as natural supporters – a more conservative older and settled cohort usually shorthanded as middle Ireland. …
“One of the main things I have been telling people is just be personable and tell your story,” says Carey. “A lot of people being canvassed have not met gay people before. It is a person standing on the doorstep saying: ‘I live up the road and I’m gay and this is about me, about real local people with the same hopes and aspirations as you.’ “Making that personal connection can really change people’s opinions. It’s no longer an abstract question about human rights. There is nothing more powerful about that personal story and the reason why they are seeking a Yes vote.”
Given what’s happened, I sincerely hope the Yes campaign hasn’t been hurt by the lie LaCour seems to have perpetrated. I don’t know what his internal state of mind was like while all this was going on. Perhaps he meticulously planned the whole thing with malice aforethought and was intending to laugh all the way to Princeton. Or perhaps he was more like an undergraduate plagiarist who suddenly finds, to his horror, that his copied paper has won a university prize. It doesn’t much matter to me. But social science can have real consequences for people’s lives—even if in this case I hope it does not. At this point the polls look pretty good for the Yes side. But it would be bitter indeed if a referendum campaign on an issue like this were to founder, or even become a close-run thing, because its organizers tried in good faith to rely on the work of a fraud.