Construct Validity

Construct Validity

1. Criterion Validity

   • evaluates how closely your measure's results align with those from existing, valid measures
     • Concurrent Validity
       • distinguishing participants appropriately
         • e.g. if assessing a new measure of depression, participants should be separated into two groups: depressives and non-depressives
     • Predictive Validity
       • measure's ability to predict future outcomes
         • e.g. reading scores at 7yo's predictive power of academic performance at 17

2. Content Validity

   • all aspects of the construct are being assessed
     • e.g. if construct pertains to personality, does the measure assess the 5 main traits? (OCEAN)

3. Face Validity

   • the measure's content seems suitable
     • e.g. does a test with math questions appear to be a relevant measure of arithmetic ability?

4. Discriminant (Divergent) Validity

   • shows two measures that are not supposed to be related are in fact, unrelated
     • correlation coefficient can be used to prove dissociation

5. Convergent Validity

   • two similar measures of the construct are significantly correlated (r= >.5, ideally >.7)
     • e.g. studying exam anxiety (construct) via a survey and participant observation. Convergence or correlation of scores shows both are measuring the same construct