Predictive validity
The extent to which performance on a selection method actually predicts performance on the job, expressed as a correlation.
Last reviewed
In practice
This is the question every hiring method should be able to answer and most cannot. Decades of personnel research have measured it across methods, and the consistent finding is that structured interviews, work samples and cognitive ability tests predict considerably better than unstructured interviews, years of experience or education level.
The absolute numbers are modest even for the best methods, and anyone quoting a very high correlation is either measuring something narrow or selling something. Hiring is a prediction problem with a lot of irreducible noise in it.
The practical implication is comparative rather than absolute. You will not find a method that predicts job performance reliably. You can straightforwardly stop using the ones known to predict worst.
Related
Structured interview
An interview in which every candidate is asked the same predetermined questions and their answers are scored against a rubric defined in advance.
Work-sample test
A selection method in which a candidate completes a bounded, realistic task drawn from the job, and the output is scored against predefined criteria.
Skills-based hiring
Selecting candidates on demonstrated ability to do the work, rather than on proxies for it such as degrees, job titles or previous employers.
Read further
Longer pieces on this and the arguments around it.
Does your assessment predict anything? Validity for non-specialists
An assessment that feels rigorous can predict nothing. What validity and reliability mean, how to check yours, and the four ways hiring tests quietly break.
ReadStructured interviews: the highest-return change most teams never make
Same questions, same order, same rubric, scored independently. Structured interviews are unglamorous, uncomfortable to adopt, and the best value in hiring.
Read