How Gender Bias Works (and how to fix it, sort of)

Gender equality
Image by duncan c,

I mentioned awhile back the sexist comments made by Olympic commentators.  Those comments are indicators of their gender bias, and they perpetuate gender bias in their audience.  It might seem petty to remark on those, but they are part of a larger picture of gender bias that permeates everything we do, not just in our sports-watching activities.  In the case of the Olympic commentary, those comments might deter young women from pursuing high levels of achievement in their sport.  And it belittles the achievements of the women their commenting on, which can reflect not just on the activity at hand but all activity by women, such that people begin to think that women can’t achieve or that their achievements are not the result of their own hard work.

Over the weekend, I tweeted this CNET article that showed that tech companies are biased when it comes to selecting candidates to interview.  You’re surprised, aren’t you? (hint: you shouldn’t be)  But you might be surprised by how big and impact the bias has.  A recruiting firm gave the resumes of 5,000 candidates to companies.  The first time, they included name, background, etc.  Of the candidates that the companies selected for interviews, 5% were women.  The second time, the recruiting firm removing identifying information and left only their job experience and skills.  This time, 54% of the candidates selected were women.  We’re not talking an incremental improvement here, but a giant increase.  This is not the first time these kinds of experiments have been done.  Researchers have changed gender information on resumes and given people names that sound hispanic or stereotypically black.  They show time and time again that given the same experience and skills, when people believe the candidate is not white and male, they’re less likely to be selected.

One person asked when I tweeted this, what happens during the interview?  The study did not go that far, but I happened to have just listened to a podcast that dealt with that very issue.  In the Freakonomics podcast, Stephen Dubner examines gender barriers.  The issue with resumes is covered, but so is the interview process. There’s a lot of good stuff in there.  I highly recommend the whole thing.  But for the interview process, one of Dubner’s guests recommends constructing a set of interview questions ahead of time, asking those questions of every candidate, and then comparing answers in a way where you don’t know who said what.  You give each candidate a number, for example.  You can then rank the answers to each question for the candidates and then come up with a total.  It’s an attempt to rid yourself of bias, whatever that may be.

We’re learning a lot about bias and about how that bias manifests itself, and then coming up with ways to short circuit that, which will, perhaps, move us toward eliminating the bias, or decreasing it.