Ask HN: What does it take to hire for merit, and avoid biases? | Hacker News Reader

Ask HN: What does it take to hire for merit, and avoid biases?

11 pointskedargj10 comments

Many companies and team leaders state that they always hire for quality, first, yet the hiring process starts with a resume even for junior positions.

I find it hard to look at a resume and not create associations to university, gpa, name, courses, etc which are not directly connected to merit of the candidate for the job.

What techniques do you use to avoid a filtering bias before conducting technical interviews?

View on HN

I'm more concerned about a diversity in hiring channels than biases on an individual level. Many startups want 'the best people' but only hire based on a candidate pool already familiar to them: friends and recent alumni of their school.

This is especially problematic when you consider that most startups are created by white males who have had a societally positive bias all their lives. And they tend to associate mostly with people similar to them. This leads unintentional to sexism, racism, and ageism in our industry, and prevents many startups from even talking to a large number of talented people.

I'd encourage you to reach out to older alumni, minority coding groups, and more experienced developers through previous jobs or your board and advisors. More diverse hiring channels will lead to a more diverse candidate pool.

1)Whatever metric you are looking to determine merit, whether it is University affiliation, GPA, or performance on a coding evaluation, don't think of the result as a single number. Think of it as a range. Even a probability distribution. The mean value of technical merit may be higher but the variance is so large that the metric may be meaningless unless you have a large number of applicants.

2)Whatever metric you are looking at, recognize ahead of time that you might be getting it wrong. So you might want to include a sample of candidates from the rest of the pile just to evaluate your evaluation methods. Unless you are the type of company getting dozens of applications from people at top schools then this shouldn't be a problem.

3)Hiring somebody right out of college is a roll of the dice anyway. No matter what their ability, you have no idea how well they work with other people, how well they listen to instructions, or how well they can handle the grind of the 9-5. Once you get to the interview phase I would look more for maturity than technical merit so long as they meet a minimum acceptable criteria. But I also suppose this depends on what stage your company is at.

I'm sorry, but how are University/GPA/Courses not directly tied to median candidate merit?

If you had someone in front of you who went to MIT and had a 4.0, and always took the most stringent professors for each course in the curriculum, how on earth do you think they compare to someone from ASU, or some community college, with a 2.15 GPA?

They aren't directly tied to merit. They may be indirectly tied to merit; that's the argument many people make. But we also know college attendance, especially at elite schools, and GPA are correlated with other things... things like economic background.

You also have to consider how bias can be self-reinforcing. We know people with stereotypical African-American names are less likely to be hired then people with the same resume but names associated with Caucasians. And black children are much more likely to be suspended or expelled then white counterparts, and to get lower grades, even if their performance and behavior is on par with white students. So credentials like a degree at a good school, and good grades, are not always free of bias.

If you are an employer, you have to consider a number of factors. You may need to fill the position immediately. You may have a certain pay scale in mind without flexibility. You want to avoid hiring a weak or bad employee, and you want to avoid not hiring a good employee who merely looks bad on paper.

In science, we often use randomized samples. You might try taking a random sample of the people who you wouldn't ordinarily pick, the no pile, and interviewing them at technical skills. Or better yet, having somebody who doesn't know who is who interview them purely on technical skills, and not letting them see the resume.

"You want to avoid hiring a weak or bad employee, and you want to avoid not hiring a good employee who merely looks bad on paper."

Yes, but the first is a much more serious problem. It's unlikely that a small company will founder if it misses a good hire, but a bad hire can certainly destroy it. Better to err on the side of caution.

If you are a manager in web hiring people based on how well they did in school/college, you are going to have a lot of issues. As someone else posted, not even Google adopts this kind of hiring approach themselves and they arguably have some of the smartest minds in the business working for them.

Some of the smartest minds in web/of our time are actually college dropouts/never attended who have never completed formal education in their lives. The only true way you become a great developer is to teach yourself, hustle and continuously learn. You can study the basics, but an evolving industry such as IT/web development can not give you all of the skills needed for the future.

I have seen many developers formally educated with all kinds of degrees from degrees in mathematics to computer science absolutely struggle in a proper work environment. Just because you can succeed academically does not mean you will succeed professionally, two different worlds with two different kinds of expectations.

I have regularly seen uneducated developers absolutely wipe the floor with educated developers more times than I can count throughout my career. Perceptions are one thing, reality is another.

Every time this thread and meme comes up I ask that posters tell us whether or not they have completed a degree and if so what degree and where. This is because I find that perceived ability usually depends greatly on how backgrounds correlate between the poster and the person you are posting about it.

I've asked this before because I don't get the American fetish for brand name universities, but never gotten a good answer. Who has shown that a degree from MIT would be better than a completely equal degree from another university? Has anyone (recently) sampled X MIT students with X other random students with the same score and shown the MIT students to be superior?

And how would MIT compare to non-American brand name technical institutes? Moscow, St. Petersburg, Berlin, Tokyo, Beijing, Bangalore ...?

If you haven't seen any evidence, but still believe that MIT is the best, how can that not be bias?

For all my fellow Illini, U of I > MIT so I've replaced them in these examples. ;)

Market signals act as a vetting process:

  * If BigCo hires a University of Illinois grad and they do well, they will continue to hire U of I grads.
  * If all your competitors are hiring U of I grads, then you'll look into hiring as well.  You can assume they too are going after only the best people.

If you believe in efficient markets, then marketplace hiring signals are all you need to choose the best candidates. This only works in aggregate of course, every specific hire is unique. Problems arise when:

  * You limit your hiring solely to a single institution that's done well in the past.  Think of it like A/B testing- if you don't give other universities a shot then you'll never know when their candidates improve.
  * The brand name prestige overshadows actual candidate ability.  Then the university becomes an old boys club, where hires make it just because they came from the right brand.

Both of these scenarios can lead to broad biases in the marketplace, due to those same signals. Your poor hiring practices send out false positives to other companies. But on an individual level most companies will respond quickly to changes in candidate ability from different universities.

This was mostly a thought experiment, I look forward to critiques.

Google HR disagrees with the the assumption about GPA correlating to ability

http://www.businessinsider.com/how-google-hires-people-2013-...

Reply on news.ycombinator.com