Poor Predictors: Job Interviews Are Useless and Unfair
In ten years, I see myself living in a world without job interviews.
Posted Aug 31, 2020
The job interview is a prominent feature of the American work landscape. We’ve all sat through one of those where-do-you-see-yourself-in-five-years conversations with a prospective employer, or employee. Ostensibly, such job interviews are designed to help select the best candidates for the job, the ones most likely to succeed. Alas, as a means of predicting future performance and success, the traditional, unstructured, tell-me-about-yourself job Interview is a hopelessly deficient and potentially harmful procedure.
The researcher Jason Dana of Yale University tells the following illustrative story: “In 1979… the Texas Legislature required the University of Texas Medical School at Houston to increase its incoming class size by 50 students late in the season. The additional 50 students that the school admitted had reached the interview phase of the application process but initially, following their interviews, were rejected. A team of researchers later found that these students did just as well as their other classmates in terms of attrition, academic performance, clinical performance… and honors earned. The judgment of the interviewers, in other words, added nothing of relevance to the admissions process.”
The traditional interview is a poor predictor of performance for multiple reasons. First, predicting the future is difficult to do under most circumstances. Moreover, an interview constitutes a tiny sample of the interviewee’s behavior. A small behavioral sample tells us little about overall behavior. That’s why you don’t get married after the first date.
Our ability to judge someone’s character, personality, or skill by a short conversation is poor. Research has shown that interviewers can’t even tell when the interviewee provides random answers. The task is rendered even more difficult by the fact that most people lie in job interviews.
Further, the qualities that make for a good interviewee have little in common with the qualities the job demands. To wit: Extroverts do better in interviews. Yet extraversion is not what most jobs require. Interviews, in other words, lack what psychologists call "ecological validity," eliciting behavior that does not reflect reliably real-world behavior.
This problem is not limited to job interviews, of course. Results in the lab may not translate well to real-world situations. Those who do well in practice may choke during the game. Our current political system is badly hampered by the fact that the skills and qualities that help one get elected to office are vastly different than those required to execute the demands of an elected office competently and fairly.
In addition, classic work by the great cognitive psychologists Amos Tversky and Daniel Kahneman (well-told in Michael Lewis’s book, The Undoing Project), has shown that the human brain contains several biases that hinder our judgment and decision making systematically. Without appropriate external guardrails—which the unstructured interview format does not provide—our mind veers off the lane of reason and rationality and into the ditch of distortion. For example, we will rely too heavily on the first piece of information offered; search for and interpret information in the way that confirms our preexisting beliefs; systematically weigh short-term future outcomes more heavily than long term ones; underestimate the cost and duration of any undertaking; and favor those who are physically attractive and people we like, regardless of their actual capabilities. People we like tend to be those who resemble us.
In addition to not doing much good, the traditional interview may in fact do harm, for example by inducing interviewer over-confidence, by perpetuating an unfair and inefficient system, and by undercutting the development of—and contribution from—more reliable sources of data. Yet it remains popular. Why is that?
For one, unstructured interviews are easy to organize and conduct and people tend to like them better than other, more structured, data-driven, and less personal approaches. Like objects under Newton’s first law, popular social customs long in motion will tend to remain in motion.
In addition, the time gap between science and practice, between discovery and acceptance, is often wide. Scientific findings are not easily disseminated, particularly when they go against something we like. Moreover, as demonstrated years ago by the aforementioned Tversky and Kahneman, we find it hard to correct our biases even after we’ve been made aware of them. The human brain is quite trenchant this way. This is one reason why placebo medicine still works even if patients know they are taking a placebo. As with memory, our confidence in our intuitive judgment tends to be both too high and poorly related to accuracy. We are quite enamored with our intuition, and the value of an interview appears intuitive—“Of course I’ll have a better sense of this person after meeting them.”
Finally, when a certain ritual persists despite failing to serve its stated purpose, it is prudent to search for an alternative, unstated reasons that may account for its perseverance. For example, an enjoyable unstructured interview experience may strengthen the interviewee’s commitment to the hiring process. An interview may not help you select the best applicant, but it may increase the odds that the applicant you’ve selected will accept your offer.
In addition, interviewers may rely on the interview as a means of uncovering some important personal information about the interviewee, such as some extreme feature of appearance or personality that may constitute a deal-breaker (or maker). Alas, the social impact of this interview function may be quite devious. After all, there are certain traits that mean a lot in our culture that are difficult to ascertain in any context other than a face-to-face encounter. Chief among those, of course, are race and "tribe." The unstructured interview may be an easy way to ascertain whether the interviewee is "one of us." Thus, it may select not for competence but for identity, keeping "them" out and letting "us" in.
The accumulated data call for us to ditch the inefficient, unfair traditional interview process. Better, more predictive and fairer options can be developed, and some already exist, among them the structured interview, which at least assures that everyone is asked the same questions; situational interview, which requires candidates to react to work-related scenarios; or "job auditions" which can be "blind," as when a musician auditions for the local symphony by playing behind a curtain.
In the meantime, my advice to employers would be: hire the candidate who refuses to submit to an unstructured interview.
LinkedIn image: Dean Drobot/Shutterstock. Facebook image: fizkes/Shutterstock