Employers, like most people, tend to trust their intuitions. But when employers decide whom to hire, they trust those intuitions far more than they should.
Suppose that you are considering two candidates for a job in sales, Candidate A and Candidate B, and have interviewed both. You and your colleagues were far more impressed with Candidate A, who was dynamic, engaging, and immensely likable — a natural, especially for sales. By contrast, Candidate B was a bit awkward and reserved, and so seemed to be an inferior “fit.”
One of your colleagues points out that both candidates have taken an aptitude test that relates to the job; their personnel files also contain their scores on a general intelligence test. On both tests, Candidate A was just OK; Candidate B performed superbly.
Which applicant will you choose? If you are like a lot of people, the answer is still Candidate A. After all, you met both in person, and part of your job is to be able to assess people. Maybe Candidate B tests well, but Candidate A knocked your socks off.
A lot of evidence suggests that in cases of this kind, employers will stubbornly trust their intuitions — and are badly mistaken to do so. Specific aptitude tests turn out to be highly predictive of performance in sales, and general intelligence tests are almost as good. Interviews are far less useful at telling you who will succeed.
What’s true for sales positions is also true more generally. Unstructured interviews have been found to have surprisingly little value in a variety of areas. For medical school interviews, for example, they appear to have no predictive power at all: in terms of academic or clinical performance, those accepted on the basis of interviews do no better than those who are rejected. In law schools, my own experience is that faculties emphasize how aspiring law professors do in one-on-one interviews — which usually provide no information at all about how they will do as teachers or researchers.
In the abstract, most people in human resources are fully aware that objective measures are helpful. Yet the overwhelming majority of people in these positions believe that executives “can learn more from an informal discussion with job candidates” and that it is possible to “read between the lines” to see whether a candidate would do well in the job. In general, that’s wrong.
In fact, some evidence suggests that interviews are far worse than wasteful: By drawing employers’ attention to irrelevant information, they can produce inferior decisions. For example, people make better predictions about student performance if they are given access to objective background information, such as grades and test scores — and prevented from conducting interviews entirely. (In some fields, of course, specific aptitude tests don’t exist, but general intelligence scores are often available. And if candidates have a previous track record, it makes sense to rely on it.)
So why do employers, managers and administrators continue to give so much weight to interviews? The simple answer is that people trust what they see and hear, and rely on their own feelings even when they shouldn’t. But as Yale University management professor Jason Dana and his collaborators have shown, there’s more to it than that. Interviewers actively fool themselves, finding ways to learn from interviews even if there’s actually nothing there to learn from.
Dana’s central finding is that interviewers work very hard to make sense of whatever interviewees end up saying. If you are conducting an interview, you will quickly form an initial impression of the candidate, and you will be inclined to assess his or her answers — whatever they are — in a way that fits with that initial impression.
To confirm that point, Dana instructed interviewees to give literally random answers to questions — answers that had nothing at all to do with their natural response. Even then, interviewers said in post-interview surveys that they received valuable information.
Dana’s explanation was that interviewers had made sense of the answers they got by weaving those answers into a coherent (and to some degree fabricated) narrative about candidates. In other words, interviewers, thinking that they are good judges of people, ended up confident about the usefulness of the interviews even when the responses were deliberately worthless.
There’s a related problem with interviews: They can give effect to biases, conscious or unconscious. If interviewers are prejudiced against women or Hispanics, for example, a face-to-face interview will predictably result in discrimination. Reliance on tests, or on actual or past performance, can promote equality.
For business, government, and education, the lesson is clear: People ought to be relying far more on objective information and far less on interviews. They might even want to think about scaling back or canceling interviews altogether. They’ll save a lot of time — and make better decisions.