In science, the ultimate measure of academic worth is the number of papers published where you’re credited as an author. There are subtleties that matter—where you are in the list of authors and whether others cite your publications. But it’s hard for those factors to overcome the weight of raw numbers. Other things, like grants and promotions, also matter a great deal. But success in those areas often depends on a large publication list.
That’s why a publication released on Wednesday by Nature is significant: It describes data that indicate that women are systematically left off the list of authors of scientific publications. The gap between participation and publication continues even after various factors of career advancement are considered. And it goes a long way toward explaining why science has a problem called a “leaky pipeline,” where women drop out of research at higher rates at each stage of their careers.
Making the team
It’s pretty easy to crunch the data and see that women are underrepresented in author lists attached to scientific papers. But figuring out why is a significant challenge. It could result from women being historically underrepresented in some fields, discrimination, or differences in effort and commitment. Figuring out which factor(s) contribute is challenging because it involves identifying an invisible population: the people who should be on the author list but aren’t.
Complicating matters is that there are no clear rules regarding what sort of contributions are needed to receive authorship. Members of a lab often help each other informally, and there’s no clear boundary where that sort of help rises to the point where it demands authorship. As a result, a large amount of politics goes into who ends up on the author list, and often lots of bad feelings among those who don’t make the cut.
If you ask a scientist about their publication history, they’ll invariably have a story about a paper they should have been credited for but were left off.
The big challenge facing the researchers behind the new paper is figuring out how to discriminate between the equivalent of office politics and the existence of widespread bias. The key bit of enabling data comes from the Institute for Research on Innovation and Science at the University of Michigan, which gathers data on more than 100 campuses that are part of 36 research universities. (For example, the University of California system is one university, but it has nine campuses, including UCLA, UC Santa Cruz, and UC Berkeley.) This data includes every grant held by researchers, any employees that grant money supports, and their job titles.
The data allowed researchers to identify 128,859 people who were part of nearly 10,000 individual research teams. Those names were then cross-referenced to databases of scientific publications, linking individuals to nearly 40,000 papers and more than 7,500 patents. This data collection allowed researchers to address a more focused question: If a scientific team is publishing successfully, are there any patterns to which team members are authors of those publications?
Who’s not there?
The researchers used several measures of publication success. The first is whether a team member appeared in a publication. Based on raw numbers, this produced an obvious answer: 21 percent of men ended up as authors, while only 12 percent of women did. But it was also clear that faculty publish far more often than anyone else on the team, and women get faculty positions at a lower rate, which explains a big chunk of that difference. So, the researchers adjusted for position (faculty, grad student, lab tech, etc.) and found that women are about 5 percent less likely to end up with their name on any publication.
Separately, the researchers figured out how many papers and patents a given team published in the period they had data for. These were considered “potential publications” for anyone who’d been in the lab a year earlier. The person’s actual publications were then divided by these potential publications to adjust for the fact that some labs or projects are more successful than others. Again, once adjusted for position, women ended up in publications at a lower rate than men.
Finally, the researchers looked into how often people end up on highly cited papers. For papers that bomb, there was no difference; women and men ended up on papers with zero citations at equal rates. But for a reasonably successful paper (one that gets cited 25 times), women are about 20 percent less likely than men to end up on the author list.
Separately, the researchers surveyed academics, getting more than 2,500 responses. Forty-three percent of the women who responded said they were left off the author list of a paper they had contributed to; in contrast, 38 percent of men were left off. Forty-nine percent of women also said their scientific contributions to the lab had been underestimated, while under 40 percent of men felt that way. Women tended to say that their work wasn’t recognized for personal reasons or because they had left the lab before the publication came out.
Finally, follow-up interviews emphasized that the lack of clear rules for authorship and the heavy role of politics contributed to the problems. People felt they had to do some self-promotion to get credit for their work. At least one individual recounted how a faculty member criticized her in front of the research group after asking for recognition. There was a consensus that these problems happened with all underrepresented groups and weren’t limited to gender.
There are a couple of noted limitations of this study. The database for tracking team contributions only contains data from a limited number of large research universities, so it doesn’t fully capture the academic research experience. The results only captured responses from people contacted via their role as research authors—people who never get onto papers weren’t contacted.
Still, the researchers did a clever job of figuring out how to identify people who were heavily involved in the research endeavor but may not have ended up getting a publication out of it. They’ve also made their data available, so people can potentially examine it for issues beyond gender.