Knowledge base Updated 2 November 2026 1 views

Construct Validity

The degree to which a measure actually captures the theoretical construct it is intended to capture, established by an accumulating argument rather than by any single coefficient.

Definition

Cronbach and Meehl introduced construct validity for the case where there is no criterion to validate against — which in psychology is most cases.

The evidence usually assembled:

Content: do the items cover the construct's domain.
Convergent: does the measure relate to measures it should relate to.
Discriminant: does it fail to relate to measures it should not.
Criterion: does it predict a relevant outcome, concurrently or later.
Structural: does the internal factor structure match the theory.
Consequential: what happens when the measure is used, which Messick argued belongs inside validity rather than outside it.

Validity is a property of an interpretation of scores for a purpose, not of a test. The same instrument can be valid for one use and invalid for another.

History

Cronbach and Meehl's 1955 paper introduced the nomological network: a construct is defined by its lawful relations to other constructs and to observables, and validation is the process of establishing that network.

Messick's unified framework in 1989 folded the older subtypes into a single argument-based account, which is the standard position in measurement theory and the one applied psychology has largely failed to adopt.

Theory

The uncomfortable circularity is intrinsic: validating a measure requires a theory, and testing the theory requires a valid measure. Cronbach and Meehl accepted this and argued the network is bootstrapped by accumulation, not resolved in one study.

Jingle and jangle fallacies are the practical failures:

Jingle: assuming two measures with the same name measure the same thing. Grit and conscientiousness scales share a name-level distinction and correlate around .8.
Jangle: assuming two measures with different names measure different things. Much of the positive-psychology literature is affected.

Construct proliferation follows: new labels are rewarded, and the field accumulates near-identical constructs with separate literatures that do not cite each other.

Research

Flake, Pek and Hehman found that a large majority of papers in a sample of social and personality journals provided little or no validity evidence for their measures, and that many scales were modified without revalidation.

Specific cases are well documented. Grit's incremental validity over conscientiousness is small in meta-analysis. Many measures of implicit attitudes have weak predictive validity for behaviour and poor test-retest reliability, which is a validity problem as much as a reliability one.

The consequence for the wider literature is direct: a substantial number of published effects concern relationships between two instruments whose construct validity was never established.

Applications

Scale selection: prefer instruments with published validity evidence in a population like yours, and check what the evidence actually shows rather than that the manual asserts it.

High-stakes use — selection, diagnosis, forensic assessment — requires validity evidence for that specific use, and the consequences of the use are part of the validity question.

As a reviewer or reader, the question worth asking of any paper is what evidence exists that the numbers measure what the abstract says they measure.

Criticism

The concept is broad enough to be hard to apply, and applied work responds by ignoring it.
No decisive test exists, so validity claims are always partly rhetorical.
Consequential validity is contested — some argue social consequences of test use are a policy matter and not a measurement one.
Practice has not caught up with theory. Messick's framework is 35 years old and most published research still reports validity as a coefficient.
The incentives reward new constructs, not the unglamorous work of showing an existing one is measured well.

References

Cronbach, L. J., & Meehl, P. E. (1955). Construct validity in psychological tests. Psychological Bulletin, 52(4), 281-302.
Messick, S. (1989). Validity. In Educational Measurement (3rd ed.). Macmillan.
Flake, J. K., Pek, J., & Hehman, E. (2017). Construct validation in social and personality research. Social Psychological and Personality Science, 8(4), 370-378.
Credé, M., Tynan, M. C., & Harms, P. D. (2017). Much ado about grit: A meta-analytic synthesis. Journal of Personality and Social Psychology, 113(3), 492-511.

Written for PsychologyOrbit and reviewed before publishing. It is a summary for professionals and students, not clinical guidance, and it is not a substitute for the primary sources listed above.