Back in December, we reported on a startling study that found a 20-minute talk with a stranger can change minds about gay marriage, if that stranger is openly gay and speaks in frankly personal terms. “The impact of such a brief talk can persist for months, and spread to other members of one’s household,” we wrote. “Open-mindedness, it seems, is contagious.”
It almost sounded too good to be true. And, in fact, it was.
The paper’s lead author, Donald Green of Columbia University, retracted the study earlier this week, after discovering that his co-author, University of California-Los Angeles graduate student Michael LaCour, apparently faked some of the data.
“I am deeply embarrassed by this turn of events and apologize to the editors, reviewers, and readers of Science,” Green writes in a letter to the editors of the prestigious journal that published the original paper. His comments are posted on the site Retraction Watch.
The problematic data was discovered by two graduate students from the University of California-Berkeley, David Broockman and Joshua Kalla, who were attempting to conduct a follow-up study.
The two men quickly found themselves questioning some of the data. Their doubts were confirmed when they contacted the polling firm that purportedly performed the original study. Officials there “claimed they had no familiarity with the project,” they write.
After bringing in Yale University political scientist Peter Aronow to review what they had found, Broockman and Kalla sent their findings to Green on May 16. The next day, Green wrote back conceding there was clearly a problem. He subsequently confronted LaCour, his co-author, who “confessed to falsely describing at least some of the details of the data collection,” according to Broockman and Kalla’s timeline of the fast-moving events.
Two days later, Green wrote to Science requesting that the paper be retracted. In it, he noted that LaCour’s graduate advisor at UCLA had asked him to supply “contact information of survey respondents so that their participation in the survey could be verified, but he declined to furnish this information.”
Green told Retraction Watch that he initially thought LaCour’s results “were so astonishing that the findings would only be credible if the study were replicated.” In response, LaCour conducted a second experiment, “and the results confirmed the original findings.”
This detail led to a pointed comment by Discover magazine’s Neuroskeptic blog, which called the incident “an important reminder that replication is not proof of integrity. If you can fake one result, you can fake two.”