Predictive validity is the extent to which a test score forecasts a future criterion. It answers the question of whether the test is useful for anticipating later performance or outcomes.

College entrance examinations are evaluated partly by how well they predict first-year grades. Employment tests are evaluated by how well they predict job performance or training success. The correlation between the test score and the later criterion is the usual index of predictive validity. A modest correlation can still be useful when the test is applied to large groups and when the costs of testing are low relative to the costs of selection errors.

Predictive validity is only one form of validity. A test may predict well yet fail to cover the intended content domain, or it may measure a construct different from the one claimed. Establishing predictive validity requires longitudinal data and careful choice of the criterion measure. When the criterion itself is flawed, the apparent validity of the test is limited.

A test with predictive validity gives a useful estimate of what is likely to happen later.