It’s not news that women are much less likely to get hired for jobs than men, even when the candidates have the exact same qualifications. Now, new research sheds light on why this happens.
Employers favor men not because they are prejudiced against women, but because they have the perception that men perform better on average at certain tasks, according to the forthcoming research paper “Is Gender Discrimination About Gender?”
The paper was written by Katherine B. Coffman and Christine L. Exley, both assistant professors at Harvard Business School, who teamed up with Stanford University economics professor Muriel Niederle.
“We find ample evidence of discrimination against women, as employers are significantly less likely to hire a woman compared to an equally able man,” the paper says. “This discrimination, however, does not appear to be driven by gender-specific stereotypes or animus.”
The findings may help employers train recruiters to be aware of their biases and work around them.
The two faces of discrimination
Gender discrimination clearly runs through the workplace. Women earn about 78 cents on the dollar compared to men. And in many industries, women are less likely to advance to the top of their fields. Women make up just 4.2 percent of CEOs at S&P 500 firms and 19.2 percent of board members.
“THIS DISCRIMINATION DOES NOT APPEAR TO BE DRIVEN BY GENDER-SPECIFIC STEREOTYPES OR ANIMUS”
The researchers wanted to take a closer look at the source of this gender divide, so they used online experiments to probe two types of gender discrimination:
- Statistical discrimination, which is rooted in beliefs about average gender differences in abilities or skills.
- Taste-based discrimination, which is driven by stereotypes, favoritism for one group, and a bias against another group.
“With statistical discrimination, you have certain beliefs about men versus women and what they can do, and given those beliefs, you choose the person who you think is the best person to hire. You are simply acting in a way that you think will maximize your profits,” Coffman explains. “With taste-based discrimination, you know a certain person will be productive, but you’re sacrificing that by not hiring that person. We did not find so much of that at all.”
While taste-based discrimination was not at play, they did find in their experiments that statistical discrimination does indeed work against women in the hiring process.
Testing for gender bias
To simulate a real-life hiring situation, the researchers created online experiments with 100 participants representing workers seeking jobs, and another 800 representing employers looking to hire workers. The workers were asked to complete a series of sports and math quizzes (stereotypically easier for men to answer), some of those questions easy and others hard. Overall, men performed slightly better than women, answering on average one extra question correctly.
Employers then had to hire a candidate, choosing between one woman and one man. Each candidate’s score results on the easy questions were made available to the hiring official, but employers were not provided workers’ scores on the difficult questions–yet they were additionally told they would receive compensation if their hire did well on the hard quiz.
When told that men did slightly better on average than women on sports or math tasks, employers were much less likely to hire a female worker than a male worker, even when two individual workers had identical easy quiz grades.
The researchers then took gender out of the hiring decision. Workers were simply identified to potential employers as either born in an even month or an odd month. (In reality, but unknown to the employers, the researchers labeled all women candidates as odd-month, and all men as even-month.) Using test results as their guide, employees still steered clear of the odd-month, or female workers, choosing them only 37 percent of the time. When identified as women, they were chosen 43 percent of the time.
“Just like the woman was hired less often, the odd-month worker was hired less often, too,” Coffman says. “That tells us the discrimination isn’t based on a prejudice against women, so it’s not that people in this setting don’t like hiring women. Instead, employees are drawing on the information about average performance and are not hiring members of lower-performing groups.”
Women are more likely than men to hire other women
Researchers also found evidence of “in-group” favoritism and “out-group” bias, meaning that employers were more willing to hire a member from the lower-performing group if the employers shared the same gender or birth month.
In the gender experiment, female employers were much more likely to hire women than male employers were. When a woman was making the decision, women were hired 50 percent of the time, yet when a male employer was making the call, women had only a 40 percent chance of getting hired. This was true with the birth month groups, too: Even-month employers were much more likely to hire even-month workers than odd-month employers were. In fact, when birth month was the consideration, rather than gender, the difference was even bigger, with odd-month employers hiring even-month workers only 30 percent of the time.
Clearly, sharing the same social identity can have an impact on hiring choices.
“It seems to be the case that all employer types, on average, are willing to engage in discrimination against members of the lower-performing group,” the paper says. “But the extent of this discrimination is reduced when the employer shares a known demographic characteristic with the lower-performing group.”
Coffman, who has conducted other research exploring the role of gender, hopes these findings will spur business executives to take a closer look at whether those doing the hiring within their organizations have general beliefs about men versus women that might affect their decisions about job candidates.
”STATISTICAL DISCRIMINATION … IS THORNIER AND IS PARTICULARLY DIFFICULT TO ROOT OUT”
“We can all agree that taste-based discrimination is a really bad thing if you’re prejudiced against women and don’t want to hire them,” she says. “But statistical discrimination, where people act on their beliefs about average differences in ability between men and women, is thornier and is particularly difficult to root out. The people doing the hiring might not even realize they are acting on those beliefs. Having discussions about what beliefs we hold could help us to understand what factors are shaping our hiring decisions, and whether we are comfortable with those factors playing a role.”
Job candidates should be aware that employers may have preconceived ideas about average ability differences among men and women in certain areas, so applicants need to provide any information they can to outweigh certain beliefs employers may have.
“Anytime an employer has beliefs about differences on average between two groups and you’re a member of that lower-performing group, that may impact your ability to be hired, even when your own individual information is strong,” Coffman says. “In our sample, the employers have a small amount of information about this person. I would hope if an employer had more information about individuals, there would be less of a need to rely on average differences, and the information about group performance should become less and less important.”