Skip to content

Stockertim Dictionary · Human resources · Technology

Predictive validity

Also called criterion validity, criterion-related validity

In short

How well scores from a selection method relate to later job performance, usually measured as a correlation between the two.

  • Advanced
  • Ratio or metric
  • Checked

In detail

Predictive validity answers a simple question: do people who score well in this assessment go on to perform well in the job? To measure it, an organisation collects selection scores, waits until hires have been in role long enough to be assessed, then compares scores with a performance measure such as manager ratings, sales or quality data. The strength of the relationship is usually expressed as a correlation coefficient between minus one and one.

In selection, coefficients are usually modest. A value around 0.5 is considered strong for a single method, and values near zero mean the method adds little. Combining methods, such as a structured interview with a work sample, can raise overall validity when each measures something different. Validity is specific to the job and the performance measure; a test that predicts success in one role may not in another.

Measuring validity well is hard. Only hired candidates have performance data, which narrows the range of scores, and performance ratings can themselves be biased. Small samples give unstable results. Organisations therefore combine their own data with published research and check that methods are fair across groups as well as predictive.

Formula

Validity coefficient = correlation(selection score, later performance)

selection score
each hire's result in the assessment
later performance
the same hire's measured performance after a set period

In practice

A contact centre compares assessment scores of 300 hires with their quality scores after six months, finds a coefficient of 0.35 for a role-play but 0.05 for a personality questionnaire, and drops the questionnaire.

Watch out

  • A method can be predictive overall yet disadvantage some groups; validity and fairness must both be checked.
  • Correlation shows association, not that the method causes good performance.

Learn it properly

Official sources

Facts last checked on 3 November 2026.

Education, not advice: the Stockertim Dictionary explains what terms mean. It never recommends buying or selling anything.

English