A lot of us are familiar with the story, thanks to economists Goldin and Rouse and later Malcolm Gladwell’s Blink, about the innovation in orchestra auditions. In the 1970s, when auditions consisted of a musician performing in front of judges, orchestras were nearly 95% male. When orchestras turned to blind auditions—ones in which the identity of the musician was hidden by a screen—women’s share of orchestras rose to about 25%. These blind auditions, it seemed, allowed judges to assess musicians on quality alone, leaving no room for gender bias (or any other prejudices) to enter the assessment process.
This leads me to consider two questions:
- Does gender bias exist outside of orchestra settings?
- If so, can blind auditions minimize gender bias outside of orchestra settings?
Let’s consider that first question. The answer is a resounding YES. The 2015 Women in the Workplace McKinsey & Company report highlights gender bias as a real factor contributing to the uneven playing field between women and men at work.
In the last few decades, social scientists have accumulated a great deal of evidence suggesting that bias of all types (those based on race, gender, age, and even place of education) influence the employment process. We have biases to help us process complex situations, make quick decisions, and navigate the world around us. But if left unchecked this bias, which Sociologist Shelley Correll calls an “error in decision-making” affects all of us (women, men, white, minority) in nearly all decisions (work-related or not).
Unfortunately, these biases shape hiring decisions in ways that hurt underrepresented groups, especially in fields where their presence has historically been very low. In other words, the biases we hold tend to lead us rely on stereotypes when making decisions, oftentimes without even us realizing we are doing so. In the world of employment, the stereotypes we hold advantage men. Our biases lead us to conclude that women do not possess the skills and attributes men stereotypically possess—intelligence, decisiveness, authority, drive which are all attributes that make for a successful worker—and we make hiring, promotion, and reward decisions that favor men. The gender bias process is especially evident in STEM fields.
I’ve summarized the findings of a few (of a growing number) of studies uncovering persistent gender bias in STEM. In the first, “How stereotypes impair women’s careers in science” (published in the Proceedings of the National Academy of Science), Ruben, Sapineza. And Zingales performed an experiment in which subjects were “hired” to correctly sum as many sets of four two-digit numbers as possible over 4 minute time period, a task women and men can do equally well. Subjects, who I will call the math performers, were paired up (either 2 women, 2 men, or a mixed sex). Some subjects were assigned to be “employers” and told to “hire” a math performer. The math performers were told they could earn more money if they were “hired” by the employer while the employers were told they could earn more if they selected the higher math performer on a future, second arithmetic task. The authors analyzed data from the male-female pairs and found that without information about performance, male and female “employers” were twice as likely to hire a man as a woman. When math performers self-report performance, men are still out-hired compared to women (possibly because men inflate their performance and women under-report it). Providing previous performance information only reduces, but does not wipe out, the male hiring advantage. The authors conclude that implicit bias results in the initial assessment of math performers and in the case when self-reported performance occurs, employers biased against women are less likely to consider that men boast more (on average) than women about their future performance. These situations could lead to employers hiring the less-skilled person for the job.
Another recent (not yet peer-reviewed) study of open-source coding by Terrell, Kofink, Middleton, Rainear, Murphy-Hill, and Parnin compares the acceptance rates of “pull requests” (proposed change to software) from female and male developers. They find that female developers have a higher pull request acceptance rate than male developers (so, no evidence of gender bias in acceptance rates) and that the differences in acceptance rates are robust to a number of alternative tests. However, when a woman’s gender is apparent, their pull requests are rejected more often than men’s. The authors conclude gender bias may play a part in their findings.
Fortunately, gender bias not unavoidable. This brings us to the second question above: can blind auditions minimize gender bias outside of orchestra settings?
That’s where the Bay Area start-up GapJumpers comes in. They have developed a system of blind auditions for entry-level workers in the tech industry. In short, the company designs “challenges,” based on the tech needs of the companies they are employed by, that they present to potential job applicants. The job applicants solve the challenges for GapJumpers who then share challenge results with companies seeking to hire. Companies judge the applicant’s challenge (without knowing anything about the applicant who submitted the challenge) rather than their resume, which contains markers of an applicant’s gender, possibly their race and social class background, and their place of education and hence, allow for assessments involving gender, race, class, or educational training bias.
I was very lucky to have Kedar Iyer, co-Founder of GapJumpers, guest speak (through Skype) to my undergraduate Sociology of Gender and Work course last week, the same day the New York Times published a story featuring GapJumpers and others confronting workplace bias. He told us that GapJumpers significantly diversifies the applicant pool of the companies they work for—companies, on average, start with a 20% diverse applicant pool and after GapJumpers, their pool is about 60% diverse.
Companies will want to know whether the applicants GapJumpers places are as successful as (or more successful?) than those hired through traditional methods. GapJumpers first report to answer this question will be out in May 2016.
Kedar also pointed out something that extends beyond the performance of GapJumper placements that companies need to pay attention to. When a job applicant from an underrepresented group is not hired by a company while their similarly skilled white, male, Ivy League degree holding friends are, the rejected applicant views the rejection as an intentional act on part of a company and spreads the word about the company’s actions. Yet, as social scientists have shown, most bias happens without intent. Nonetheless, unintentional bias is interpreted as negative treatment and could tarnish company reputations, so controlling unintentional bias in the way GapJumpers has been able to do may be key to survival in a competitive field.
Readers of this blog may be wondering how blind auditions can be incorporated into other industry settings—if not into the hiring process, into other systems of evaluation. Others might be asking whether blind auditions will really work—that is, are interpersonal attributes, and things like “cultural match” or “emotional capital” discovered only through face-to-face meetings necessary if not to hire the best, but to ensure the applicant “fits”?
As for me, it seems one way companies, universities, schools, and other organizations can convince the public they do more than lip service to diversify their ranks is to go the way of orchestras and GapJumpers and as best they can, make blind auditions standard practice.