Scientists pride themselves on their objectivity, yet when it comes to gender and race, they are as partial as everyone else. A 1999 study, for example, found that academic psychologists were more likely to recommend hiring a male job applicant than a female one with an identical record. A résumé assigned the name Brad Baker is more likely to lead to a job interview than one from Rasheed Jones. Numerous studies have shown that women scientists get weaker letters of recommendation. Whereas female applicants to a faculty position at a medical school were called “teachers” and “students,” men were “researchers” and “professionals.”
These kinds of biases are unconscious and subtle but invidious enough to suppress the diversity of students and faculty in many university science, engineering and math departments and in the scientific workforce at large. As of 2010, white men made up 51 percent of all working scientists and engineers, in contrast to white women (18 percent) and black and Hispanic men and women, who each held 4 percent or fewer of these jobs. As Katherine W. Phillips shows in “How Diversity Works,” starting on page 42, diversity is not only socially just, it is an essential ingredient in high-quality scientific work.
Asking individuals to check their own predispositions is a worthy step, but it is insufficient. Unconscious biases cannot be wished away. Institutions must strive to eliminate opportunities for implicit bias to affect decisions on hiring and promotions. Systemic changes are crucial.
One simple way to help ensure that rewards go to the most deserving applicants—not just the ones with the right names—is to strip critical documents of identifying information. Hiring committees cannot favor white and male applicants if résumés have only a number at the top. Peer reviewers cannot disproportionately reject journal papers from women and ethnic minorities if author identities are hidden.
A 2012 study of Swedish data on real-life job applications showed that anonymous hiring practices increased the chances that women and minorities made it to the interview stage. The same thing happened in a 2010–2011 German pilot program when participating companies removed personal details such as age and gender from job applications. Symphony orchestras have found that they are more likely to hire women when musicians audition from behind a screen.
University science departments can and should obscure the identities of applicants on curricula vitae (CVs) and other materials during the first round of screening for new faculty members and graduate students. That won't be a panacea, of course: in some circumstances, the identity of an applicant can be sussed out from the details and collaborators on the CV. Still, “anonymization” introduces an element of doubt that helps reviewers strive for objectivity. By the time a candidate shows up for the interview, the hiring board will have had a better chance of forming a first impression on the basis of work and experience, with no stereotypes attached. Anonymity can also help in grading classwork and reviewing grants. It should be mandatory in universities and grant-giving institutions.
Keeping authors of scientific papers anonymous has been shown to improve women's odds of acceptance. The practice can also block bias against minorities and in favor of authors and institutions with big reputations. Double-blind peer review, in which the identities of both authors and reviewers are hidden from one another, is already common in the social sciences and humanities, and now science journals are starting to pick it up. A trial of double-blind peer review began in June 2013 at Nature Geoscience and Nature Climate Change, and Conservation Biology is considering adopting the practice. (Scientific American is part of Nature Publishing Group.) More science journals should follow their example.
Language is also important. When a National Institutes of Health award emphasized “aggressive risk taking” among its criteria, all nine of the first winners were men; later, terms such as “pioneering” and “high impact” attracted and rewarded more women. Training in identifying and compensating for implicit bias should be mandatory in scientific organizations. And because training is never perfect, documents under review should be anonymous. Awards must be based on merit—not a name.