Skip to main content



Gender bias found in recommendation letters

06 Oct 2016
Photograph showing someone choosing a candidate from a selection of photographs
Implicit bias: study finds researchers write stronger recommendation letters for men. (Courtesy: iStock/AndreyPopov)

Female postdoctoral fellowship applicants are half as likely as their male counterparts to receive glowing recommendation letters, according to a study by researchers at Columbia University’s Lamont-Doherty Earth Observatory (LDEO). Led by Kuheli Dutt, assistant director of academic affairs and diversity at the observatory, the researchers also found that both male and female scientists tend to write stronger recommendation letters for men than for women. The findings add more evidence of implicit, or unconscious, bias that women are perceived as weaker in the sciences than men.

While previous studies have shown that those who evaluate applications for postdoctoral fellowships or other scientific positions tend to have implicit bias against women, the new study focuses on the other side of that application – the person writing a recommendation letter. The new analysis focused on geoscience research positions – and specifically postdoctoral-fellowship applications for positions at the LDEO. Dutt and colleagues analysed 1224 recommendation letters – redacted of personal information – that were written between 2007 and 2012 by 1101 researchers from 54 countries.

The researchers looked at the tone of each overall letter and classified it as either “excellent”, “good” or “doubtful”. Terms like “outstanding”, “genius” and “groundbreaking research” were classified as excellent letters, for example. To compile a “tone-terminology manual”, the researchers spoke to 18 senior scientists who either currently serve on or have in the past served on postdoctoral-fellowship selection committees.

Meaningful dialogue

Out of the 1224 letters, 862 were written for male applicants and 362 for female applicants. The researchers found that for male applicants, 24% were rated as excellent, 73% good and 3% doubtful. For letters for female applicants, 15% were excellent, 83% good and 2% doubtful. “We’re not assigning blame on anyone,” says Dutt. “Everyone has implicit biases. It’s more about using these results, and the fact that they’re consistent with implicit bias, to engage in meaningful dialogue or to open up conversations that would lead to rectifying the problem.”

Dutt and her colleagues acknowledge that they were unable to statistically rule out the possibility that male applicants may have been better qualified than females. “That being said, [the results] fit with the literature,” says physicist Zahra Hazari from Florida International University. “This is one of many studies that have found things of this ilk.”

Reflecting on beliefs

According to Hazari, who is a member of the American Physical Society’s Committee on the Status of Women in Physics and was not involved in the study, the findings “amplify the issue of gender bias” because it comes from both sides of the application process – from the recommender and the evaluator. “We can’t change the stereotypes or the views overnight,” says Hazari. “But what we can do is reflect on our own beliefs and how we’ve internalized the beliefs that are out there. By thinking about them, we can neutralize some of the effect.”

The work is published in Nature Geoscience.

Copyright © 2021 by IOP Publishing Ltd and individual contributors