Is Hiring for Culture Fit Another Form of Unconscious Bias?

Two recent hiring trends have enjoyed growing popularity among big organizations.

The first is to attempt to minimize–if not abolish–implicit or unconscious biases in their recruitment process, in particular, during the job interview. For example, many diversity and inclusion programs offer unconscious bias training to managers in order to make them aware of their prejudices toward candidates from minority groups. Famous examples include Starbucks, Vodafone, and Facebook, but there are many others.

The second is to hire for “culture fit,” which is to ensure that external recruits match the profile of successful employees, so they can easily adapt to the new role and the organization as if they already belonged. While this is typically evaluated in terms of candidates’ values, style, or background, such qualities are generally folded into demographic and socioeconomic factors. Prominent examples include Google’s focus on Googliness, Bridgewater’s ruthless radical transparency, and Airbnb’s hipster culture of belonging.

But what if hiring for culture fit actually decreases diversity? Is hiring for fit just another form of unconscious bias? Or perhaps even a conscious bias, since it represents an explicit selection criterion that employers are proud to stand behind?

There are three factors that suggest the answer might be yes (even though it’s unfair to generalize).

Subjectively judging someone’s potential based on intuition

In an age of abundant alternatives for making data-driven evaluations of talent and potential–including established scientific assessments and digital talent technologies–there’s no excuse for playing it by ear. Yet the most common method for evaluating a candidate’s potential is the unstructured job interview, which is a weak predictor of future job performance. The interview is especially used to assess culture fit. At worst, it boils down to a gut feeling of good chemistry or rapport that interviewers get from the candidate. At best, this results in well-meaning interviewers trying to ignore the very factors that cause that experience, such as charisma, attractiveness, and likability, as well as any attributes or background they share with the candidate. Unlike AI or machine-learning algorithms, it’s simply not possible to train humans to “un-see” candidates’ gender, age, or race, or “un-learn” these categories. In fact, the more we try to suppress something, the more present it is in our minds.

Unreliable measures of performance

For most employees, and especially those in professional jobs, performance is typically assessed by a single rating handed down by the employees’ direct supervisor. In general, this person has limited objective data to qualify his or her rating, which tends to be contaminated by politics, likability, and personal preference. That same manager tasked with the performance review was likely also responsible for hiring the employee. This means they have every incentive to show that they were right to choose that candidate and give them a high rating. As Alan Kay famously noted: “The best way to predict the future is to invent it.” Interviews may seem predictive of future performance, but they are simply reinforcing the hiring manager’s bias, which then functions as a sort of self-fulfilling prophecy of culture fit.

If organizations want to mitigate this bias, they should make an effort to get independent, objective, and reliable measures of job performance without subjective perceptions or managerial preferences. Think of how Uber evaluates its drivers. The app tracks all the key indicators of performance, such as how many trips drivers make, how much money they make, how many trips they cancel, whether they have any accidents, and their customers’ satisfaction. This leaves very little room for subjectivity, and no humans are needed to rank different drivers on the basis of their performance. Although such precision may not be feasible in most jobs, organizations should still try to approximate this level of objectivity.

A demographically homogeneous representation of high performers

The more similar a company’s high performers are in terms of age, gender, race, experience, and technical expertise, the more likely it is that hiring for culture fit will reduce diversity. There’s a reason why culture and cult have the same root. A strong culture inhibits both demographic and cognitive diversity. It’s always easier to manage a large group of individuals when they are all the same. This is the ultimate excuse for getting away with bias when hiring for culture fit: the inability to create an inclusive culture and leverage the benefits of people who are able to think differently.

That said, culture fit doesn’t have to lead to a lack of diversity. But in order to achieve this, it’s essential that organizations rely less on the job interview, and more on objective predictors of job performance, that they measure job performance objectively, and that they have cultures in place that enable people with diverse backgrounds to perform highly at work.