Better hiring decisions

During a job interview, have you ever come across – either as candidate or interviewer – questions such as ‘How many golf balls would fit inside a 747?’ or ‘Estimate how many petrol stations there are in Beijing’? Interviewers in the early days at Google were renowned for asking similarly idiosyncratic questions.

Later analyses by the technology firm established that these sorts of questions did not predict subsequent employee performance. Laszlo Bock, senior vice president of people operations at Google, admitted nearly a decade ago that such questions ‘have little if any ability to predict how candidates will perform in a job’.

Data from both private and public sector organisations shows that most untrained interviewers are typically poor at recruiting candidates who go on to become high performers. When I train managers in how to interview candidates, I find that around eight out of 10 believe their interviewing skills are above average. Statistically, this is impossible: in reality, many managers overestimate their ability to evaluate candidates.

Structuring the interview

Traditional, unstructured interviews allow interviewers to ask whatever questions they wish. In contrast, structured interviews offer guidance to interviewers on both questions to ask and how to evaluate candidates.

A large body of evidence has confirmed that interviewers in structured interviews tend to recruit candidates who go on to add more value to their organisations. A study by Wei-Chi Tsai at National Chengchi University in Taiwan found that candidates' non-verbal behaviour – their body language – affected interviewers' hiring decisions far more in less structured than structured interviews.

In unstructured interviews, candidates who present a more vivacious version of themselves end up with better ratings on unrelated skills; candidates who smile and use more gestures are likely to be perceived as having more commercial judgment or analytical skills, for example. In other words, it's easier for candidates to dupe interviewers in unstructured than structured interviews.

In terms of structuring interview questions, both past-based questions (‘Tell me about a time you dealt with a difficult client’) and hypothetical questions (‘How would you deal with a difficult client?’) have validity in helping interviewers to judge likely future job performance. Past-based questions tend to measure relevant experience. Hypothetical questions may be more indicative of general mental ability or intelligence.

Past-based questions and hypothetical questions both predict future performance reasonably well for simple jobs. However, past-based questions are a better predictor of performance in complex managerial jobs. This suggests that managerial performance is better predicted by job-relevant experience than raw intelligence.

Other types of questions exist. For example, a study led by Utah State University's Christopher Hartwell found that both past-based questions and so-called background questions (eg ‘What experience have you had of giving formal presentations?’) predicted subsequent employee turnover. Hypothetical questions did not predict employee turnover.

The superiority of structured over unstructured interviews has been proven by an analysis of a large data set of correlations between interview scores and subsequent job performance ratings. Frank Schmidt and Ryan Zimmerman at the University of Iowa found that one structured interview by a single interviewer was as predictive of future job performance as three to four unstructured interviews by separate interviewers.

Clearly, having one structured interview saves time – especially as the three to four separate interviewers must also spend time discussing each candidate to reach their decisions. However, interview structure needs to be developed beforehand; there is an upfront investment of time and perhaps consulting fees in choosing the most effective interview questions and rating scales.

Selecting strong performers

Surveys have shown that candidates perceive structured interviews to be more difficult than unstructured ones. But if the purpose is to filter out weak candidates without the qualities to do the job, perhaps the greater difficulty of structured interviews may be a good thing.

Structured interviews also reduce bias against minority groups, making them legally more defensible, too. To create a rudimentary interview structure:

1. Get hiring managers to discuss the key skills or qualities that differentiate high performers from average ones in the workplace.
2. Based on those skills and qualities, decide on broad questions to ask all candidates for any given role.
3. Ask the recruiting managers to agree scoring criteria for candidates. What kind of evidence would justify a high versus a low score on each skill or quality?
4. Finally, train interviewers to use the questions and scoring criteria in a broadly consistent fashion.