Construct Validity

Construct validity asks whether an instrument actually measures the abstract concept it claims to measure. Because constructs like motivation, trust or service quality cannot be observed directly, they are inferred from indicators — and construct validity is the evidence that the inference is sound.

It is established through two complementary strands. Convergent validity shows that indicators of the same construct converge, evidenced by standardised loadings above about 0.70 and AVE at or above 0.50. Discriminant validity shows the construct is distinct from others, evidenced by HTMT below 0.85 or the Fornell-Larcker criterion.

Threats include construct underrepresentation, where the measure captures only part of the concept, and construct-irrelevant variance, where it picks up something else such as social desirability or common method bias. Adapting a validated scale and pilot-testing it is far safer than writing items from scratch.