In what has been called “gaygenegate” in some corners of the Internet, a conference presentation last week about the genetics of homosexuality in men has come under intense scrutiny. The talk also raised questions about whether scientists working on controversial topics should post unreviewed preprints of their findings before presenting them at a meeting. Statistician Andrew Gelman of Columbia University in New York, who criticized the homosexuality study’s statistical analysis, wrote in a blog post that the lack of a peer-reviewed paper or preprint made it difficult for people to evaluate the work. Other researchers countered that conferences are meant to be forums for early, unpublished work.
During a 15-minute talk on 8 October at the American Society of Human Genetics (ASHG) conference in Baltimore, Maryland, data scientist and geneticist Tuck Ngun of the University of California, Los Angeles, discussed his work on twins that found associations between certain epigenetic changes, or chemical alterations to DNA, and being gay. But the study’s small sample size, statistical methods and conclusions, as well as the conference's press release and the way the news media covered the story, all came under attack.
In his blog post, Gelman wrote that there were statistical errors in Ngun’s analysis. He added that he would normally explain this kind of “statistical misunderstanding” as errors that get past peer review in the scientific publication process. “But … in this case there is no paper! No publication, not even a preprint,” he notes.
“Preprint allows third parties to verify or comment in a way that’s harder to do here,” Gelman said in an interview. “I don’t think there’s anything wrong with someone giving a talk and not having a preprint, but it is very hard to judge .”
Robert Lanfear, a biologist at Macquarie University in Sydney, Australia, responded to Gelman’s blog post on Twitter:
In an e-mail, Lanfear said that conferences are supposed to be a place to talk about early work, but a preprint would be a good idea in cases in which the results are bound to attract a lot of attention. “If pressed, I would say that press releases without at least a preprint are a recipe for disaster.”
Other scientists said that asking for preprints at conferences might prevent researchers from presenting in the first place. “Most people don’t pre-publish anything when they’re giving a ten-minute talk,” says Jim Woodgett, a stem-cell researcher at the Lunenfeld–Tanenbaum Research Institute in Toronto, Canada, who also commented on Gelman’s blog post. “You really don’t want to discourage people from talking about work that is maybe not complete.”
Still, he added, researchers should be prepared to discuss details of their experiments and statistical analyses after the talk. Otherwise, “You do leave yourself open to significant criticism,” he says.
Preprints are routine in the physical sciences but have been a source of continuing debate online among life scientists. Structural biologist Stephen Curry at Imperial College London wrote in a blog post in September that he supports them as a supplement to the long process of peer review.
Biologist Michael Eisen of the University of California, Berkeley, went further earlier this month and pledged to abandon pre-publication peer review altogether and post his papers online for public review. Other researchers have also chatted on Twitter about whether preprints are citeable, and whether they establish priority in a competitive field.
Ngun issued a rebuttal to the criticism of his work, defending his statistical analysis and saying: “I have always been clear that it is too early to determine the nature of the relationship of the genes to sexual orientation.” In an e-mail, he said that the ASHG conference did not ask for a preprint. “In my experience, preprints are extremely rare in the life sciences. I have not been to a conference that has requested one.”
For more, see www.nature.com/socialselection
- Journal name:
- Date published: