We need a more nuanced approach to predicting job performance.
Editor’s note: This article is part of a new MIT SMR series about people analytics.
Have you ever taken an aptitude or work personality test? Maybe it was part of a job application, one of the many ways your prospective employer tried to figure out whether you were the right fit. Or perhaps you took it for a leadership development program, at an offsite team-building retreat, or as a quiz in a best-selling business book. Regardless of the circumstances, the hope was probably more or less the same: that a brief test would unlock deep insight into who you are and how you work, which in turn would lead you to a perfect-match job and heretofore unseen leaps in your productivity, people skills, and all-around potential.
How’s that working out for you and your organization?
My guess is that results have been mixed at best. On the one hand, a good psychometric test can easily outperform a résumé scan and interview at predicting job performance and retention. The most recent review of a century’s worth of research on selection methods, for example, found that tests of general mental ability (intelligence) are the best available predictors of job performance, especially when paired with an integrity test. Yet, assessing candidates’ and employees’ potential presents significant challenges. We’ll look at some of them here.
People Metrics Are Hard to Get Right
For all the promise these techniques hold, it’s difficult to measure something as complex as a person for several reasons:
Not all assessments pass the sniff test. Multiple valid and reliable personality tests have been carefully calibrated to measure one or more character traits that predict important work and life outcomes. But countless other tests offer little more than what some scholars call “pseudo-profound bullshit” — the results sound inspiring and meaningful, but they bear little resemblance to any objective truth.
People often differ more from themselves than they do from one another. Traditional psychological assessments are usually designed to help figure out whether people who are more or less something (fill in the blank: intelligent, extraverted, gritty, what have you), on average, do better on whatever outcomes the organization or researcher is most interested in. In other words, they’re meant to capture differences among people. But