Skip to main content

Table 3 Generic hypotheses used for the assessment of construct validity and responsiveness

From: Measuring carer quality of life in Duchenne muscular dystrophy: a systematic review of the reliability and validity of self-report instruments using COSMIN

A priori rules for hypothesis testing
Convergent validity
 1 If construct being measured is the same or similar to that measured by the instrument then the correlation should be r ≥ 0.5
 2 If construct being measured is dissimilar but related to that measured by the instrument then the correlation should be between r = 0.3 and 0.49
 3 If construct being measured is judged as unrelated to that measured by the instrument then the correlation should be r < 0.3
Known groups validity/responsiveness
 1 Differences between groups where a large difference is expected should be d ≥ 0.8
 2 Differences between groups where a medium difference is expected should be between d = 0.5 and 0.79
 3 Differences between groups where a small difference is expected should be between d = 0.2 and 0.49
 4 Differences where no or a trivial difference is expected should be d < 0.2