Views on bias can be biased

The scientific method is supposed to be objective. A scientist comes up with a hypothesis for how they think something works. Rigorous testing takes place. The scientist looks at the resulting data objectively, and determines whether the original idea was right or wrong. There’s no arguing with numbers. When in doubt, test it again. As my graduate adviser always said, “the data are the data.”

But that doesn’t mean that science is bias-free. After all, science is an endeavor performed by scientists — people who come with their own beliefs and biases that influence how they see the world. A new study of gender bias in academic research puts the lens not on that bias, but instead on the interpretation of studies of it. And the results show another level of bias: Male scientists view studies of gender bias in academic research less favorably than do female scientists or male researchers in other fields. But the good news is that combating preconceived notions about gender bias research can turn faculty hiring toward equality.

When Jessi Smith, a social psychologist a Montana State University in Bozeman, received a grant from the National Science Foundation to increase the participation of women in science, she started meeting with faculty to tell them about gender bias in academic research. As part of the meetings, she handed out copies of a study by Corinne Moss-Racusin and colleagues at Princeton University in New Jersey published in 2012 in the Proceedings of the National Academy of Sciences. The study found that male and female science faculty ranked job applicants as more hirable if their applications featured male names — a subtle gender bias in favor of men.

“Their article showed men and women hold these biases to the same degree, no blaming here, not just one gender at fault,” Smith says. But she soon realized that not everyone saw the study the same way. Many of the men she passed the paper to “started picking apart the study design and the flaws they saw with the paper.” Smith often ended up trying to convince the scientists that there were many other papers out there showing gender bias in science. “I was very surprised,” she recalls. “I thought it was going to seal the deal. You can’t argue with data.” But it turned out that many scientists could, and did, argue with the data.

Ian Handley, a social psychologist also at Montana State, overheard Smith discussing the issue in the department hallway. “I was really interested,” he says. As he and Smith discussed her observations, they wondered if, when it comes to studies of gender bias, “people just don’t want to hear it.” They decided to conduct a survey to find out whether scientists really do approach research about gender bias with some preconceived notions.

Using Amazon’s Mechanical Turk, the scientists recruited 205 people (146 men and 59 women). The participants were asked to read the Moss-Racusin abstract and rank how favorably they viewed the study. Questions included how well participants felt the abstract was written, if they thought the research was important, and if they agreed with the results. While both men and women viewed the abstract relatively favorably, men ranked it significantly less favorably than women.

Those people were not scientists.  So the researchers also surveyed 205 academics (those in science, technology, engineering and math, as well as those in non-STEM fields) from Montana State. They presented participants with the same abstract and questions. Again, while both genders ranked the abstract generally positively, men ranked the abstract less favorably— agreeing with the results less, ranking the study as less important and less well-written — than women did. When the faculty were divided into STEM and non-STEM disciplines, non-STEM faculty showed no gender differences in their rankings. Only in STEM fields did the male faculty rank the abstract less favorably.

In a final test, the researchers went back to Mechanical Turk, and recruited another 303 people. These people received a second abstract by Silvia Knobloch-Westerwick and colleagues at Ohio State University in Columbus. This paper, published in Science Communication in 2013, showed that people rank science conference abstracts more favorably if they are associated with male authors. This time, the researchers presented either the original abstract or an abstract that altered the findings to show no gender bias in conference abstract ranking. Men once again gave the abstract showing gender bias less positive rankings than women did. But when the paper was altered to show no bias, male views changed, and they liked the abstract more than the real, bias-demonstrating results. Smith and colleagues report their findings October 12 in Proceedings of National Academy of Sciences.

Why does the bias exist? Handley notes the effect could be a matter of confirmation bias. “Women already believe there’s bias,” he says. “It fits their experience and they evaluate [the abstracts] favorably. Men, they don’t see it every day and don’t think it’s a problem.” It could also be a matter of defensiveness. After all, Handley explains, most men don’t want to think they are bad people who are biased against women.

And most men, of course, aren’t bad people, any more than most women. “They’re not saying that male faculty in STEM are evil or trying to put people down,” says Sarah Jane Leslie, a philosopher at Princeton University. “These are implicit biases and everyone has them. It doesn’t make you a bad person, it makes you a human being.” She says the study itself is fascinating and important. “We need to acknowledge these issues so we can talk about them,” she explains. “If some people are less likely to acknowledge [these issues], it’s very detrimental.”

Part of the problem, explains Erica Cantrell Dawson, a social psychologist at Cornell University, is that “this is one area where people use their own personal experience to decide whether [research] is legitimate or not.” She says it’s natural to read a scientific paper and “check in, ask, ‘is this how I experience the world?’ And if men are not experiencing the world that way, they will be skeptical.” Is it a biased view? Sure. “It’s not the way science should be done,” Dawson notes. “But it’s the way human reasoning goes.”

The most surprising thing, says Ellen Peters, a psychologist who studies decision-making at Ohio State, is that no one had done this study before. “As humans, we’re influenced by a huge number of influences and biases,” she says. “If these people are being biased against women and it can be explained by very subtle, unintentional differences, that’s holding back a whole world of people contributing to our society.” Peters says she would like to see the study replicated with more participants, as well as to see just how much these preconceived notions might contribute to behaviors such as hiring.

Smith and her colleagues have already started applying the results of their research. They designed an intervention, randomly assigning hiring committees to a program meant to help people recognize unintentional bias and give them tactics and resources to successfully recruit more gender-diverse faculty.  And the intervention has already worked: Hiring committees who got the intervention were 6.3 times more likely to offer a position to a woman than committees that got no intervention. The improvements went the other way as well. When they got offers from committees that received education about combating gender bias, women were 5.8 times more likely to accept, the researchers report October 10 in Bioscience.

So preconceived notions of bias do exist. But tackling the problem can result in improvement. “Everyone is going to be better if we address this inequality,” he notes. “The more ideas we bring to the table, the better our science is going to be.”