Types of validity evidence
| Type of validity evidence | What is it? |
|---|---|
| Face | A judgement that, on the face of it, the test is likely to measure the intended construct13 14 |
| Content | A judgement that the test adequately represents the construct of interest. COSMIN requires a judgement about relevance and comprehensiveness of the item13 14 |
| Criterion | The extent to which test scores correlate with scores on a ‘gold standard’ test of the same construct. Not applicable to PROMs of constructs for which no ‘gold standard’ test exists. COSMIN considers that the term criterion validity in testing of PROMs applies only when testing a short-form version of a PROM against the original version14 |
| Predictive | The extent to which test scores accurately predict a future outcome13 |
| Construct validity | The extent to which the test measures the intended construct and the inferences that can therefore be made from the scores13 |
| Convergent | The extent to which test scores correlate with scores on another (concurrently administered) test of the same construct13 |
| Discriminant | The extent to which test scores do not correlate with scores on another (concurrently administered) test of a different construct13 |
| Known or extreme groups | The extent to which test scores differ between groups of persons known or expected to differ on the variable of interest13 |
| Responsiveness (longitudinal validity) | The extent to which change in test scores reflects change in the construct13 14 |
| Structural | In the COSMIN terminology, this refers to the examination of the extent to which all the items in a PROM reflect a single construct (unidimensionality), or the extent to which sub-scales can be demonstrated (multidimensionality)14 |
| Cross-cultural validation | The process of translation and retesting the measurement properties of a test in another language/cultural setting14 |
COSMIN, consensus-based standards for the selection of health-status measurement instruments; PROM, patient-reported outcome measures.