March 19, 2014, 6:00 AM —
Image credit: flickr/Joselito Tagarao
Coding tests or evaluations are a routine part of hiring software developers these days. But how much do personal biases of the interviewers affect the final evaluation of a candidate's coding skills? Do interviewers subconsciously (or otherwise) give, say, women or older candidates lower scores for their coding just because of their gender or age? If so, what can be done about it?
While there's no question that women are underrepresented in programming and that older programmers are more scarce, it's hard to say how much of that is due to bias on the part of those doing the hiring. It's not hard to imagine, though, that such bias is present to some degree in the tech world. If so, programmer Jason Gorman may have hit on a (partial) solution to the problem: blind programmer coding evaluations.
Gorman wrote this week about his recent experience conducting programmer evaluations using double blind pair programming. Concerned about about interviewer bias, Gorman decided to run blind coding tests of six candidates he was asked to evaluate for a client. By "blind" Gorman meant that he would not meet the candidates in person or speak to them directly (to avoid hearing their voices) or even read their CVs. All he knew about about each candidate was that he or she claimed to know Java.
For the evaluation, he "met" with each candidate via IM on Skype and only discussed the task at hand. He logged into each candidate's screen via TeamViewer and them gave him or her Java code and asked the candidate to refactor it to eliminate as many code smells as possible in 30 minutes. In the end, two of the candidates "went offline" after getting the instructions, two others muddled along unimpressively and the remaining two did well enough to be recommended for the job. What happened from there, he didn't yet know.
Assuming there is some sort of hiring bias against certain groups in programming based on sex, age or race, would this kind of blind evaluation help? It might. As someone on Hacker News pointed out, this is reminiscent of a study done by researchers at the National Bureau of Economic Research which found that once symphony orchestras moved to blind auditions in the 70s and 80s, the percentage of females in the top five orchestras in the U.S. jumped from 5% to 25%.
I think blind evaluations for programmers is a very intriguing idea well worth doing. Of course, it wouldn't eliminate interview bias completely, since somebody at the company with a say in who gets hired is, obviously, going to see the candidate's CV and meet with him or her in person. But I suspect that blind coding evaluations could at least help reduce or eliminate bias when it comes to evaluating a candidate's coding abilities, which could only be a good thing.
Read more of Phil Johnson's #Tech blog and follow the latest IT news at ITworld. Follow Phil on Twitter at @itwphiljohnson. For the latest IT news, analysis and how-tos, follow ITworld on Twitter and Facebook.