Since reading Ibram X Kendi's How to be an Antiracist, I'm starting to re-examine certain policies I've taken for granted. What I've previously thought of as meritocratic or race-neutral might be neither; it might instead be part of the problem.
In that book, he gives a clear criteria for whether a policy or idea is a racist one: Does it establish or reinforce racial inequality?
With that in mind, I thought I'd look at my own house -- the tech industry -- and at our very real tendency to run companies composed mostly of white males.
There are many reasons why this happens, but I'd like to drill into just one: The university degree requirement.
Most "good jobs" these days require some sort of university degree. Tech goes one step further, and asks for a degree specifically in computer science or another STEM field.
The degree isn't enough to get the job, of course. Most interview processes still test skill level at some point. But the field of candidates is narrowed, deliberately, via this requirement.
The question is: Does requiring this technical degree bias the selection process towards White people?
Before diving into the statistics, let's back up and talk about the criteria here. How can we tell if the degree requirement biases selection?
In order to do that, we need to know what an unbiased selection process would look like.
And here is where it's important to note the composition of the general US population (and why the Census being accurate is so very very important). If all things are equal between racial groups, then the composition of Congress, company boards, and job candidates will reflect their percentages in the population.
Anything else is inequality between the races, and can only be explained in one of two ways: either you believe there are fundamental differences between people in different racial groups (which, I will point out, is a racist idea), or there are policies in place which are creating the different outcomes.
With that criteria established, we can examine the possible racial bias of requiring university degrees by looking at two numbers:
- How many people of each racial group obtain STEM degrees in the United States?
- How does that compare to their level in the general population?
Who Has a Degree, Anyway?
According to 2018 data from the US Census, approximately 52 million people (out of a total US population of 350 million) have a bachelor's degree in the US.
Of those 51 million, 40.8 million are White.
Only 4.7 million are Black.
That means White people hold 79% of all the bachelor degrees, while Black people hold only 9%.
Their shares of the general population? 76.3% White, 13.4% Black.
So Whites are overrepresented in the group of people with bachelor degrees, and Blacks are underrepresented.
So by requiring any university degree, at all, we've already tilted the scales against Black candidates.
Who is Getting Degrees?
But what about new graduates? Maybe the above numbers are skewed by previous racial biases in university admissions (which definitely happened), and if we look at new grads -- those entering the workforce -- the percentages are better?
I'm sorry, but nope. If anything, it's worse.
Let's drill down to just those getting STEM degrees (since those are the degrees that would qualify you for most tech jobs). In 2015, according to the NSF, 60.5% of STEM degrees were awarded to White people, and only 8.7% of them went to Black people.
The same report notes that the percentage of degrees awarded to Black people (~9%) has been constant for the last twenty years.
So universities, far from leveling the racial playing field, actually reinforce inequality.
Simply by asking for a university degree, then, we're narrowing our field of candidates, and skewing the talent pool we draw from so that White people are overrepresented.
Thus, we're more likely to select a White candidate, simply because more White people are able to apply.
That reinforces racial inequality, and makes requiring a university degree for a job -- any job -- a racist policy.
What can we do instead? To be honest, if your current interview process can't tell candidates who have the right skills from candidates who don't, then requiring a college degree won't fix it.
If your interview process leans heavily on discovering a candidate's background, instead of their skills, re-balance it. Come up with ways to measure the skills of a candidate that do not require disclosure of their background.
In programming, we have all sorts of possible skill-measuring techniques: Asking for code samples, having candidates think through a problem solution during the interview, inviting essay answers to questions that are open-ended but can only be completed by someone with engineering chops.
By asking for a demonstration of skill, rather than personal history, we'd both make our interviews better -- because we'd be filtering for candidates who have shown they can do the job -- and less biased.
And if we're serious about increasing diversity in our workplaces, we'll drop the degree requirement.