Use people analytics to uncover unconscious bias in hiring decisions

Full disclosure: I have biases. Some of them I'm aware of, and inevitably some of them I'm not aware of.

News flash: you have biases too. Some of yours may be the same as mine, and some are likely different. And if your reaction to my declaration was to think "I'm not biased," here's another news flash: you're deluding yourself, and in addition to whatever your biases are, you're also suffering from the bias blind spot .

Unconscious Bias

As a focus on Diversity, Equity, & Inclusion continues to take on a more prominent role at many companies, one element is commonly found at many companies. Increasing the focus on unconscious bias is a tentpole of many DE&I strategies.

If you've made your way to this article, I'm going to assume that you're already familiar with the concept of unconscious bias. In the chance that you're not, there are many other resources that can better educate you than I, so recommend you turn to them first. But to put us all on the same page, as a common definition let's understand unconscious - or implicit - bias to be "the various social stereotypes and judgments that people unknowingly assign to others based on a variety of factors, such as their age, socioeconomic status, weight, gender, race, or sexual orientation."

Unconscious bias is a recognized issue that can have negative impacts in the workplace. As such, many companies seek to educate their employees and help them see their own biases, in hopes of eventually eliminating or reducing them. Some companies develop their own training in-house, but many turn to external vendors for training.

There's nothing inherently wrong with unconscious bias training. It's a necessary first step: you can't address any issue if there isn't a common understanding of what it is. But the problem many companies haven't solved is that unconscious bias training by itself is a necessary but not sufficient intervention. Meaning that it must be done for almost any other intervention to have an impact, but by itself, is not enough to solve the problem.

Research backs this up. Recent analyses cited by Harvard Business Review showed that "unconscious bias training did not change biased behavior." Even worse, other studies revealed it can actually make the problem worse. "Training can backfire. Sending the message that biases are involuntary and widespread—beyond our control, in other words—can make people feel they’re unavoidable and lead to more discrimination, not less."

So what's one to do?

Turn towards data

Once a company is beyond the training component of unconscious bias and has everyone on the same page as to what it is and why it's important to focus on, they can begin to turn their energies towards rooting it out and eliminating it. This is where people analytics can step in and significantly help.

As an example, let's look at gender representation in hiring. Imagine a company that already has a 50/50 mix of men and women employees, and in any given year hires a 50/50 mix of men and women as new employees. Viewed in a simple visual, their external hires' gender diversity would look something like this:

One interpretation of this might be that the company has achieved - and is maintaining - gender parity. But any even mildly seasoned people analytics practitioner would very quickly exercise some healthy data skepticism and know that this surface level observation was likely hiding some disparities underneath it, and would turn to data to dig into it further. For example, is there gender parity in their hires by job level, organizational hierarchy, or some other dimension?

But let's keep our focus on the gender parity at the top of company, and simply break down by the hiring manager. Specifically by the gender of the hiring manager. Years of established research has already shown that we tend to identify with and like people who are like us, creating a fertile ground that unconscious bias could creep into a hiring decision.

When breaking the exact same external hires down by both their gender and the gender of their hiring managers, a particular pattern may very possibly emerge. The visual below, to some degree or another, is not uncommon to see at many companies:

As you can see, in this example, men tend to hire men instead of women at a 3:1 ratio, while women hire women at a 3:1 ratio.

Now if you were to go to any one of those hiring managers, and ask if they felt that unconscious bias might have played a role in their hiring decisions, my prediction is that nearly all of them would say they didn't think so. Some might even reference the unconscious bias training they previously took as evidence that they couldn't be biased...

For any company to truly assess the presence of unconscious bias and its impact on their hiring decisions, they need to do a little more analysis. Other factors may be affecting their data, such as the gender mix of qualified applicants or the types of roles being hired for. Once controlling for those factors, a significance test can help show if the different gender hiring mix between men and women hiring managers is statistically significant or not.

Bringing it closer to home

Using these internal insights simple at the top of company level is a great way to help bring the lessons from unconscious bias training closer to home for managers. This is actually a great opportunity for a partnership between People Analytics and DE&I COEs to help make the training feel more real and relevant to a company.

But for any individual hiring manager, it's still very hard for data to reliably show a pattern of bias, because most hiring managers simply don't hire enough people in any given year for a visible trend to emerge. But as we saw, when we aggregate many hiring decisions together across multiple hiring mangers, a pattern can become visible.

So how do we get managers to see and feel their own patterns of bias? Instead of aggregating hiring decisions together across hiring managers, think about aggregating a hiring manager's decisions together across time.

Reflecting on our own unconscious biases

One useful method to open managers' eyes to their own potential biases is to have them conduct a "self-audit" of their own past hiring decisions. This can extend well beyond their tenure with the company, and use data that is unavailable in your company's people analytics platform. Use your own data to test out your assumptions of how biased you might be.

In full transparency, I'll go first.

I began my career as an officer in the military, during which I didn't exactly "hire" employees so much as have them assigned to my unit. But in the time since I left the service, I have been the decision-making hiring manager for 10 employees (bonus: this makes the math super simple!). For context, except where noted otherwise, I'll compare all to the free Visier People Benchmarks as of Q2 2022.

• Four were women, for a long-term rate of 40%. As a comparison, 49.4% of new hires (less than 1 year of tenure) today are women. If one more hire had been a woman, I would be right in line with the benchmark.
• Three were people of color, for a long-term rate of 30%. I was a little further off here, as 41.4% of new hires are people of color. This might be an area of opportunity for me.
• Two were veterans, for a long-term rate of 20%. Decently higher than the national average of about 5.6%, according to 2021 data from the Bureau of Labor Statistics . I am a veteran myself, however, so it is possible this was an unconscious bias of mine.

Now I fully understand that there are a lot of factors out of the hiring manager's control that can affect the number of diverse applicants they even receive for a job posting. Things like geographic location, industry, job function, and more can impact a woman or person of color's decision to apply for a job.

For example, for the past several years I've been in HR, which is typically over-represented by women; all three of my hires so far in HR were women. Prior to that I worked in aerospace manufacturing, which typically has a lower representation of people of color.

Conduct your own self-audit

I challenge each of you now to do your own self-audit of your past hiring decisions. You may not like what you see, but that's okay. Awareness is one of the first steps towards change. Use the insight you gain to try to reduce your bias for future hires. Use it to open up a dialogue with your DE&I leaders. And if you're brave, post a comment to this article and publicly acknowledge your biases and any insight you gained about yourself.

BONUS: Reflecting on our own unconscious biases, part two

Question: When you first viewed the two charts above, did your brain initially jump past the legend to begin interpreting the chart, implicitly assuming that the pink bar represented women and the blue bar represented men? Or worse, did you have a reaction that the chart was "wrong" because men were indicated with pink? If you did, man up (see what I did there?) and take it as a sign that you might have other, deeply ingrained unconscious biases.