Objectivity is tough, even for scientists.
Because of unconscious gender bias, scientists are twice as likely to write glowing letters of recommendation for men than for women, according to a comprehensive new study of recommendation letters in the geosciences published last week in the journal Nature Geoscience.
Surprisingly, it did not matter if a woman or a man wrote the letter itself, the bias remained.
Glowing letters included language the researchers designated as “excellent” and described the candidate as a “rising star” or “brilliant” and “superb.” Women were more likely to receive “good” letters that described them as “hardworking” and “diligent.” Yet men and women were equally as likely to get “doubtful” letters.
“This is implicit bias,” Kuheli Dutt, a co-author of the paper and the diversity chair at the Lamont-Doherty Earth Observatory at Columbia University, told The Huffington Post. “If there was conscious sexism, women would have a larger number of doubtful letters.”
The research, which focused on letters written for PhD students’ postdoctoral positions, follows several other studies of hiring in the sciences that confirm the pervasiveness of “implicit or unconscious bias,” or the notion that humans unwittingly make certain assumptions about race and gender that can unconsciously influence their actions. That scientists are plagued with implicit bias does not mean they’re sexist ― after all the bias was exhibited by both women and men ― but that their brains have developed certain associations of which they’re unaware: Men are more likely to be viewed and described as aggressive leaders and women as hard-working nurturers.
Implicit bias shows up in other ways in our culture ― in policing and education for example, where data reveal that black people may get stopped by police more often or black children suspended more frequently.
Unconscious bias is such an accepted term at this point that companies, like Google and Microsoft, are teaching employees about it in an effort to hire more women and people of color. There’s even software out there that promises to help hiring managers strip implicit bias out of job listings.
This specific research on letters of recommendation is important because these letters are critical in helping doctoral students make the leap up onto the first rung on their career ladder.
The researchers looked at 1,224 recommendation letters, submitted by recommenders from 54 countries, for postdoctoral fellowships in the geosciences over the period 2007–2012. They stripped out gender identifiers before sorting the letters into the three categories. Excellent letters were absolute raves, and said things like “X is exceptionally talented and brilliant, and has the potential to become an internationally famous scientist” and “X is a rising star of the 21st century in this field.”
The good letters make clear the candidate is solid if not a star: “X is hard-working, sincere, intelligent and very motivated” and “X has a good foundation in this field since s/he took many courses in it.”
The doubtful letters clearly show that recommenders are not supportive of the candidates, with language like “X is very set in his/her ways and does not accept advice or critique from his/her colleagues.”
The Lamont researchers focused just on the geosciences ― a male-dominated field that includes geology, oceanography and meteorology. Women earn PhD’s in this field at near the rates of men, but aren’t making it up the ranks to full professor. Women receive 40 percent of the doctoral degrees in the geosciences, but only hold 10 percent of full professorships.
But the problem with bias in recommendation letters has come up in studies in other fields. In one analysis of recommendation letters in the field of psychology, published a few years ago, men were more likely to be described in so-called “agentic” terms ― that conveyed their assertiveness, independence and self-confidence. Women were more likely to be described in communal terms ― descriptions of kindness, sympathy, helping others, acceptive direction and so on.
And implicit bias in the hiring process is not confined to the sciences by any stretch. Other researchers have looked at the interview and promotion process and performance reviews as areas that need attention.
It’s worth emphasizing again, as the researchers in the Lamont study do, that this kind of bias is not intentional. For that reason, it can be even more pernicious ― it’s hard to fix a problem of which no one is cognizant.
“We aren’t trying to assign blame or shame,” Dutt said. Hopefully the research will start a conversation and raise awareness, she said. It would be productive if, even privately, people who write these letters start looking at the language they use for men and women ― just to get a better sense of what’s going on. “It’s a larger problem.”