SCIENCE is rife with prejudice. The first ever study to compare objective measures of the quality of scientists鈥� work with the subjective ratings they are given by other researchers when applying for grants has revealed that women have to be 2.5 times more productive than men in order to get the same peer review ratings. And that is in Sweden, a country named by the UN as the world leader in sexual equality.
Christine Wenner氓s, a microbiologist, and Agnes Wold, an immunologist, both at Gothenburg University, had observed that women scientists tend to be only half as successful as men in the competition for postdoctoral fellowships from Sweden鈥檚 Medical Research Council.
The two researchers wanted to know whether female applicants were really rated objectively, so they asked to see the evaluation scores. The research council twice refused to release the data, and only after a court battle鈥揳nd with the help of Sweden鈥檚 freedom of information legislation鈥揹id Wenner氓s and Wold get their hands on the evaluation documents.
Advertisement
The researchers found that, for the 1995 intake of postdocs, the peer review committees rated women lower than men in each of three measures: women received 0.25 fewer points on average than men for scientific competence, 0.13 fewer for the relevance of proposals, and 0.17 fewer for the proposed methodology. Not surprisingly, men outnumbered women among the successful applicants by four to one.
In theory, the women could have been ranked poorly because their work was of lower quality. Wenner氓s and Wold tested for this by looking at an established measure of the impact of published work. Rather than simply taking a scientist鈥檚 total number of publications, this measure weights each paper according to the average number of times papers in the particular journal are cited by other scientists in a calendar year.
In this week鈥檚 Nature (vol 387, p 341), Wenner氓s and Wold report that women and men with the same publication impact were awarded vastly different scores for scientific competence by their peers (see Figure). 鈥淚 don鈥檛 think they were consciously discriminating,鈥� says Wenner氓s, who is now at the Pasteur Institute in Paris, 鈥渂ut there鈥檚 a tendency to over-value men鈥檚 achievements and undervalue women鈥檚.鈥�
To rule out the possibility that women happened to work in areas of low priority for the research council, or that they came from less-renowned universities, Wenner氓s and Wold ran further statistical analyses. They found that apart from publication impact and gender, the only factor that had a significant influence on the competence scores was being a colleague of a review committee member.FIG-mg20830201.jpg
Even though colleagues were not allowed to rate each other, knowing that the applicant had worked with a committee member was enough for the other reviewers to give a higher rating of competence. 鈥淚f you happen to be a female candidate with no connections, you have almost no chance,鈥� says Wenner氓s.
鈥淚f these results are true, they are worrying,鈥� says Peter Collins, a policy adviser at the Royal Society in London, who has studied the process of peer review.
鈥淚t鈥檚 the sort of thing one suspects, but to actually have it there in bold just made me angry,鈥� says Susan Greenfield, a neuroscientist at the University of Oxford. She argues that funding agencies in Britain and elsewhere should submit data on their evaluations to the same scrutiny.