Criterion Validity

The strength and direction of the association show how closely scores track the selected benchmark and whether higher test scores correspond to higher or lower criterion values. A strong association supports confidence that the assessment reflects the relevant outcome, whereas a weak association raises concerns about its usefulness. Interpretation therefore depends on both magnitude and direction, not merely whether a relationship exists.

Concurrent and predictive approaches differ mainly in timing. Concurrent validity compares assessment scores with a criterion collected at the same time, making it useful for evaluating agreement with a current benchmark. Predictive validity compares scores with an outcome measured later, so it addresses whether the assessment can anticipate subsequent performance, symptoms, personality-related behavior, or other relevant outcomes.

The criterion must be relevant to the construct and intended use because an association with an unsuitable benchmark can misrepresent a measure’s value. For example, a test intended to assess symptoms should be judged against an appropriate symptom-related outcome rather than an unrelated standard. Selecting the criterion is therefore a substantive psychological decision, not only a statistical step.

A basic evaluation begins by selecting the construct measure, identifying an established external criterion, and deciding whether the criterion will be assessed concurrently or later. Researchers then compare the test scores with that benchmark and examine the association’s strength and direction. The resulting evidence is interpreted in relation to the test’s intended purpose, such as screening, selection, or educational assessment.

Criterion validity is especially useful when a psychological assessment will inform a practical decision. In diagnostic screening, it can indicate whether symptom measures align with an external outcome; in personnel selection or educational assessment, it can inform whether scores correspond to relevant criteria. The same logic can support intervention planning by showing whether measured characteristics relate to outcomes that matter for decisions.

Weak evidence does not simply label a test as unusable; it signals that the measure may not adequately serve its intended purpose. Researchers may need to reconsider the selected criterion, the timing of comparison, or the assessment’s suitability for the target construct. In practice, this caution can prevent overconfident diagnostic, selection, educational, or intervention decisions based on poorly supported scores.