Once upon a time, if you went for a job interview, you had a fairly clear idea about the questions you’d be asked – your education, previous experience, reasons for wanting that particular role.

Not so now. The “unstructured” job interview is becoming increasingly popular. In this format, there are no standardised, predetermined questions. , it’s more like an open-ended, free-flowing conversation led by the interviewer, who decides what to ask as the interview progresses.

Employers say they prefer this approach to the structured interview because they believe they get to know candidates better. Although that may be true, there’s no evidence that unstructured interviews predict job potential.

Jason Dana and colleagues at Yale reviewed the relevant literature and conclude that unstructured interviews have little predictive validity. Why, then, they wondered, do employers continue to use them?

To answer this question, they created a series of experiments in which they asked participants to predict fellow students’ current grade point average (GPA). Some participants were given the students’ previous GPAs (“valid information”) only, while others were given GPAs and also shown transcripts from unstructured interviews with the students.

As a twist, some of those who’d had an unstructured interview were told to answer truthfully, while others were instructed to answer using a random response system.

When asked to predict GPAs, participants were less accurate when they had access to the unstructured interviews as well as students’ previous GPA’s than when they were offered GPA’s only – and this was true of students who answered truthfully as well as those who had given random answers.

When questioned later, participants admitted they relied more on interview data than on GPAs – even when they noticed that some students’ answers seemed to make little sense – so convinced were they that interviews contain important predictive information.

Even more distressing for job candidates is a newer version of the unstructured interview, the “brainteaser”, in which candidates are deliberately asked impossible questions (for example: “How many windows are there in London?” or “How many taxis are there in New York?”). The idea is that such questions gauge applicants’ reasoning skills – but, as with unstructured interviews, there’s no evidence they identify the best candidates for particular jobs.

Furthermore, according to new research carried out by Scott Highhouse at Bowling Green State University, individuals who use brainteaser questions are more likely to show a number of negative personality traits, such as narcissism – traits best avoided in an employer or colleague.

What, then, are the most reliable ways to make a wise choice if you’re hoping to hire a suitable candidate? According to Jason Dana, structured interviews, in which everyone is asked the same predetermined questions, are more accurate than unstructured interviews. Even better, however, is to rely on practical exercises that test a candidate’s relevant job-related skills.

In other words, keep it simple and straightforward. That way, companies will hire more appropriately skilled employees, and employees will be spared an unnecessary and unpleasant introduction to their new workplace.