Explainer · Plain-language

Criterion Validity: Definition, Meaning & Examples | CASRAI

Criterion validity is the extent to which a measure correlates with an external criterion — an independent benchmark of the same construct. It is usually split into concurrent validity (measured at the same time) and predictive validity (measuring a future outcome).

CASRAI plain-language explainers — clear answers to recurring research-administration questions

The step most authors miss

Doing CRediT right? Don’t stop at the statement.

A CRediT statement credits you inside one paper. The recognition CRediT was built for happens when those roles are tied to you, persistently. Sign in with your ORCID — free — and claim your CRediT contributions on casrai.org, the home of the standard. They become a verified, portable part of your identity, not a line that disappears into one PDF.

Free: claim your contributions, then export a journal-ready CRediT statement, schema.org structured data, JATS XML, CSV or BibTeX — and preview your public profile. A membership publishes that profile publicly and verifies the journals you serve.

Validating against an external benchmark

Criterion validity is established by correlating a measure with an external criterion believed to reflect the same construct. A new short depression screen might be correlated with scores from an established clinical interview; an aptitude test might be correlated with later job performance ratings. The size of the resulting correlation — the validity coefficient — summarises how well the measure tracks the benchmark. This empirical anchoring to an observable outcome is what gives criterion validity its practical force.

Concurrent versus predictive

The two main types differ in timing. Concurrent validity correlates the measure with a criterion gathered at roughly the same time — useful when you want a quicker or cheaper substitute for an established assessment. Predictive validity correlates the measure with a criterion in the future — central to selection and screening, where the point is to forecast an outcome such as university success, relapse, or job performance. The same measure can show different concurrent and predictive coefficients.

The criterion problem

Criterion validity is only as good as the criterion. If the benchmark is itself unreliable, biased, or a poor reflection of the construct, the validity coefficient is hard to interpret — a problem methodologists call the "criterion problem". Coefficients are also attenuated by unreliability in either measure and by restriction of range when only a narrow slice of people is studied (for example, only those already hired). Good criterion-validity work scrutinises the criterion as carefully as the measure being validated.

Relation to other validity types

Criterion validity is one strand of the broader validity picture. Where content validity reasons about item coverage and construct validity assembles theory-driven evidence, criterion validity supplies a direct, outcome-anchored check. Predictive validity is best treated as a specialised case of criterion validity focused on forecasting. In modern frameworks, all of these are facets of a single unified argument that a measure’s scores support the intended interpretations and uses.

Key facts

At a glance

Definition: How well a measure correlates with an external criterion
Concurrent: Criterion assessed at the same time as the measure
Predictive: Criterion is a future outcome the measure should forecast
Evidence: A validity coefficient (correlation with the criterion)
Key limit: Needs a trustworthy criterion (the "criterion problem")
Attenuators: Unreliability and restriction of range lower coefficients

Common misconceptions

What people often get wrong

Often heard: Criterion validity and construct validity are the same thing.

Actually: No — criterion validity correlates a measure with an external benchmark; construct validity assembles broader theory-driven evidence that the measure behaves as expected.

Often heard: A high validity coefficient proves the measure is sound.

Actually: No — it is only as good as the criterion. A biased or unreliable criterion makes the coefficient hard to interpret.

Often heard: Concurrent and predictive validity always give the same result.

Actually: No — they differ in timing and can yield different coefficients; predicting a future outcome is a distinct task from matching a present benchmark.

Going deeper