The structured interview is the strongest predictor of job performance. The science is settled.
For twenty-five years, the field believed cognitive ability was the best single predictor of who would succeed in a job. In 2022, a landmark re-analysis quietly changed the answer — and it has direct consequences for how you should run your first round.
If you've ever heard the claim that IQ tests predict job performance better than anything else, you were hearing the legacy of Schmidt and Hunter's 1998 meta-analysis — for decades the most-cited paper in personnel selection. It shaped how a generation of assessment vendors positioned their products and how HR leaders justified their tooling.
In 2022, Paul Sackett and colleagues revisited that foundation and found a methodological flaw running through it: the way earlier researchers corrected for "range restriction" — the statistical fact that you only observe performance data on people you actually hired — had systematically inflated some validity estimates and deflated others. When they applied more defensible corrections, the rankings shifted.
The new order
Under the updated analysis, the structured interview emerged as the strongest predictor of job performance — ahead of general mental ability. The structured interview's operational validity landed around ρ = .42, while cognitive ability, revised under more conservative corrections, came in closer to .31. That is not a rounding error. It is a reordering of the field's central hierarchy.
Structured, not just "an interview"
The word doing the heavy lifting is structured. Meta-analyses consistently show that structured interviews roughly double the predictive validity of unstructured ones — the free-flowing "let's just have a chat" format most managers default to. Structure means the same competencies, the same questions, the same scoring rubric applied to every candidate. It's the difference between a measurement and an impression.
This is also why structure reduces bias. When every candidate faces the same calibrated questions and is scored against the same rubric, there's far less room for the interviewer's mood, similarity bias, or first-impression halo to steer the outcome. Structure improves accuracy and fairness at the same time — they aren't a trade-off.
The operational problem
The best-validated selection method is also the hardest to scale. A truly structured interview demands trained interviewers, consistent rubrics, and disciplined scoring — for every candidate. Most enterprises can't staff that, so they reserve structure for finalists and screen everyone else on the resume.
Why this is suddenly actionable
Here's the tension the science exposes. The method with the best evidence behind it is the method enterprises apply least, because it's the most expensive to run consistently at volume. So the strongest predictor of performance gets rationed to the final round, while the weakest signals — the resume, the unstructured phone screen — do the heavy filtering earlier, when the pool is largest and the stakes for a false negative are highest.
What changes the math is automation that preserves structure. An AI interviewer that asks the same role-calibrated questions of every candidate, scores them against the same rubric, and documents the reasoning is, in effect, the structured interview made scalable. It moves the best-validated method from the finalist round to the first one — which is exactly where a strong predictor does the most good.
The science has been settled for a few years now. The interesting question is no longer whether structured interviews predict performance best. It's whether your process actually uses them where it counts — or whether you're still rationing your best instrument to the last five people in the funnel.
Sackett, Zhang, Berry & Lievens (2022) — Revisiting the design of selection systems, Journal of Applied Psychology
Society for Industrial and Organizational Psychology — commentary on Sackett et al.
Master-HR — Insights from Sackett et al. (2022, 2023)