Hi. This post is for information – I’m not actually hiring at the moment, but when it next happens I would really like to know how to do it right. Also this is entirely a post off my own bat, rather than an official Aframe position on anything (though I’d expect and hope the rest of the company would agree with me).
So I work with a really great development team at Aframe. We’re a bunch of competent interesting people, and everyone is well motivated and good at their job. It’s nice.
There’s just one problem…
I say “people”, but that’s because I tend to default to gender neutral nouns and pronouns. I really mean “men”.
Generally speaking, straight white men.
I’m not 100% sure how this happened.
As far as I can tell, it’s not the result of explicit prejudice on our part. I don’t think it’s even the result of implicit prejudice, and I’d like to tell you a little bit about our current hiring process in order to back that up.
Here is the process we used for our recent hiring of back end developers:
- Applicants come in to our development manager. He does a basic screen for “not an utter waste of time” (e.g. “I am definitely suitable for your senior backend developer role because I know photoshop”) then passes the candidate on to the rest of us.
- We review the answers to a set of screening questions. We do not even know the candidate’s name, let alone CV, at this point. They are given a candidate number.
- If we like their answers, we send them a coding test. Still just a number, not a free person.
- If we like their answers, we’ll take them through a phone interview. We will vaguely look at a CV at this point for context for the interview. If their answers are borderline and we’re inclined to reject we might use the CV to get more evidence that they’re worth talking to, but we never rejected anyone on the basis of their CV at this stage.
- In the phone interview we ask the same list of questions of everyone. There is of course some general chatter as well, but in theory the decision is made on the basis of the answer to those questions.
- If they pass the phone interview, we take them through to an in person interview. This is a series of specific, structured, tasks.
- We make a hire/no hire/wait and see decision based entirely on the outcome of those tasks.
I largely think this is an example of Doing It Right. It’s not perfect, and we’ve had some complaints about specific details which we’re trying to improve, but it does a pretty good job at eliminating biases.
Except the only people who made it through to the phone interview stage in the recent hires I was involved in (for which we hired three people in the end) were white men.
I honestly have no idea if this was representative of the candidates we didn’t let through to phone interview. I assume it is, but because that stage of the process is blind I don’t really have any way of checking (I’ve asked to see if we can find a list of who the candidates were, but I don’t think we collated that information). But it would be surprising if these tasks introduced a bias. The only plausible biasing factor to them is that they’re quite time consuming (we’re working to reduce that as a factor – currently considering swapping the order of the phone interview and coding test).
As partial evidence that this works in eliminating bias, the Aframe back end dev team is significantly older than you might expect than a start up tech team. At 29 I think I’m the youngest in the back end team by some margin, and I’m the second youngest overall. I think the dev team’s median age is over 40, and the mean certainly is. Given that age bias is a huge problem in technology, and that a lot of it is probably subconscious, I think this is at least a point in favour of the process if not wholly convincing. It might also be that those of us who were hiring just don’t care that much about age, so hard to say for sure.
So, my assumption is that something is happening that is filtering people out before they get to us. I have three possible candidates for this:
- Our job postings may be off putting in some way. Historically this was almost certainly the case, but I think we fixed that. I may be wrong. This is what we posted
- Recruiters. We did use job boards (I think stack overflow was our most successful one there), but I think everyone we actually hired came through a recruiter. Recruiters are reportedly often assholes to deal with if you’re a woman
- The initial “not an utter waste of time” review is not blind, and that’s a bit of a red flag. It’s supposed to be sufficiently black and white, with any shades of grey passed on to the rest of us, that I’d hope that it wasn’t introducing any significant biases here, but it was invisible to me so I can’t guarantee that
- A lot of it is probably just the background population. My impression is that “straight, white and male” is by far the dominant demographic in the London startup scene. I don’t have statistics to back that up at the moment, but either way I’m sure it’s not as total as we were seeing. So this is probably a large factor, but I can’t imagine it’s the only one.
So, that’s enough about us. Now we get to the real point of this post: What can we actually do to improve things?
Useful advice I’ve received so far:
- Get people from a range of demographics to review your job posting, even if they’re not interested in applying
- Be really explicit with recruiters that you want a diverse range of candidates. If they claim they don’t have a diverse range, use a different recruiter
- Target professional networks that have a more diverse selection of members rather than relying so heavily on recruiters
- Keep doing what we’re already doing – keeping as much of the screening process as possible blind. Where not possible, keep it structured. Where not possible, at least try to decide based on concrete results rather than “I liked them”
Like I said, we’re not hiring at the moment, but when we do next it would really be nice if we knew what to do to avoid this happening. Any advice you can offer would be greatly appreciated.