An interview in which every candidate gets the same questions and every answer a score agreed in advance predicts job performance better than any other selection method. That is the conclusion of Paul Sackett and three colleagues in a re-analysis of decades of selection research, published in the Journal of Applied Psychology in 2022. The unstructured interview, where the interviewer lets the conversation run, predicts half as well.
The figures
The researchers express predictive value as a correlation between the method and later performance. The structured interview comes out at 0.42, the unstructured interview at 0.19. A job knowledge test reaches 0.40, a work sample 0.33 and a cognitive ability test 0.31.
That is a different order than long assumed. In Schmidt and Hunter's much-cited 1998 overview, the cognitive test and the structured interview both stood at 0.51. Sackett and colleagues show that the old corrections inflated most figures. The structured interview now stands alone at the top.
Why it works
An unstructured interview mostly measures rapport. Every candidate gets different questions, and the interviewer compares impressions instead of answers. A structured interview forces a comparison on what the role requires. Two interviewers who score independently also catch each other's blind spots.
What you can do
- Pick four to six requirements that truly define the role, and write one question per requirement about something the candidate has done.
- Agree per question what a weak, an adequate and a strong answer sounds like.
- Have each interviewer score separately, and only then discuss the scores.
- Keep room for an open part at the end. It just does not count in the score.
Sources: Sackett, Zhang, Berry and Lievens, Revisiting meta-analytic estimates of validity in personnel selection, Journal of Applied Psychology 107(11), 2022; Schmidt and Hunter, Psychological Bulletin, 1998.
Pearson & Partners locates and qualifies candidates on the requirements that truly define a role.