Dumb Ways to Address Implicit Bias

For those of you who don’t know, implicit bias is the unconscious use of stereotypes in making decisions. It turns out that even if we work hard to avoid bias, our subconscious leads us to make some decisions that prioritize one group over another. For example, there’s evidence that people read identical applications differently if they identify the candidate as a man than the read it if they identify the candidate as a woman. They also read it differently if they identify the candidate as white rather than if they identify the candidate as a person of color.

So, how do you address implicit bias? From what I know about the research, there are a variety of things you can do. The first of which is to acknowledge that we all have implicit biases; it’s hard to address a problem if you don’t first acknowledge it. A second thing is to revisit any decisions you make, asking yourself whether implicit bias may have played a role [1]. As I read the literature, I see that we are also more likely to make quick judgements (and therefore incorporate implicit bias) when we are under time pressures or stress, and that we reduce implicit bias by providing more time and less stress.

I’m on a few job searches for staff positions, and the College appropriately wants to try to avoid implicit bias in those searches. We’re doing some very sensible things. For example, we’re making sure that we set criteria early on in the search, before evaluating candidates, and we’re putting an equity advocate in many searches so that there’s someone else to ask if we are perhaps incorporating implicit assumptions about the people we are evaluating.

So, you may be asking yourself, implicit bias is an important issue to address, and Sam says that Grinnell is doing some sensible things. Why did he call this dumb ways to address implicit bias? The answer is that I haven’t listed all of the things we’re doing.

Because we may make assumptions about a person based on their race and gender, we are trying to eliminate information on race and gender from applications for the first screen. (When we do phone interviews, it’s harder to avoid making assumptions about gender.) But we’re also striking any information about where they got their education [2]. And, while I acknowledge that one reacts to different institutions differently, and even that someone can be mediocre and have gone to a top school, or excellent and have gone to a mediocre school, you also learn a lot about someone from the school they attended, particularly if they are only a year or two out. For example, if I’m hiring a programmer fresh out of college, I’ll certainly look to see what their work looks like on GitHub, but I’ll also look at what their undergraduate curriculum is like. And so we lose an important piece of information.

At the same time, we don’t strike other information about their education, such as their major or GPA. I’d expect that those are potentially as biasing as the institution. Worse yet, we seem to leave in salary expectations (which is one of those questions that I think is explicitly biasing [3]) and salary history (ditto [4]). Why are we even asking these questions?

I’ll even admit that there are ways to answer these questions that will bias my responses. If you’re applying for a programming job and only want $30,000, I have to assume that you’re not a very good programmer. If you’re applying for a SysAdmin job, and you list an hourly wage, I will probably assume that you don’t understand that SysAdmin should be treated as a profession with a salary, and not an hourly job. (Yes, those are explicit biases.)

Now, Grinnell is relatively new at trying to address implicit bias in our searches (I think CS volunteered to be the first test case two years ago), and we all acknowledge that we have room to grow. But I’d hope that we’d think a bit more carefully about what people already say about appropriate approaches. I also hope that we’d acknowledge that some information, while potentially biasing, is also important.

When I was discussing this issue with someone not on the faculty, they asked how, in faculty searches, we avoid making unconscious (or conscious) choices based on institutional reputation. Now, I can’t speak for other departments, but I know that in computer science we focus very much on the statements a candidate provides. Does their teaching statement show someone who is a passionate and creative teacher? Can they explain their research in a way that a non-specialist would understand it, and does their research seem to be something that could involve students? Does their diversity statement show that they understand the issues that face our field and that they have a commitment to addressing those issues? Do they write clearly, grammatically [5], and thoughtfully?

Personally, I also work to address implicit institutional bias by regularly reminding myself that some of the top faculty at Grinnell, people who have taught and inspired me, did not necessarily attend the top institutions in their field. I also know that, although I attended one of the top overall institutions, I was in a new program, and it was not necessarily among the top programs. But I seem to have done relatively well at Grinnell, and I expect others in my situation could, too.

I do the same when someone starts to ask questions like Does this person really understand a liberal arts education? That may not sound like a biased question, but it often boils down to Did they attend a top liberal arts college? However, since the student populations at elite liberal arts colleges do not adequately reflect the diversity of our country (or the world), we will often find that such a question eliminates some worthwhile candidates. So, how do I help people address that question? I remind them that some of the faculty who we would most closely identify as being real Grinnell faculty only attended schools like the University of Virginia and the University of Pennsylvania.

Where was I? Oh, yes. I don’t think you address implicit bias by striking useful information. I think you address implicit bias by discussing it. And I now know that you address implicit bias by trying to reduce the stress on those making decisions. I hope my administrators know that last approach.

[1] Those were the first things that came to mind. As I look at some recommendations, I see that my inclinations are correct.

[2] Note that these strategies don’t necessarily apply to faculty searches. We do talk about implicit bias, and we do have equity advocates for those searches. However, I don’t think we’ve reached the point that we’re striking information from applications.

[3] There’s some evidence that majority candidates will be more likely to ask for and negotiate for higher salaries.

[4] There’s even more evidence that this practice ensures that people who take a low-paying job, for whatever reason, get stuck in low-paying jobs.

[5] Amazingly, I misspelled grammatically on my first try.

Version 1.0 of 2016-09-30.