What to Do With "Moderate" Reliability and Validity Coefficients?
editorial · Level V
Where this comes from
- Record sourced from PubMed, PMID 27095143.
- Also identified by DOI 10.1016/j.apmr.2016.04.001.
- No licence information is recorded for this record.
- Because redistribution is not established, this page shows the abstract only. Follow the links below for the full text.
Abstract
Clinimetric studies may use criteria for test-retest reliability and convergent validity such that correlation coefficients as low as .40 are supportive of reliability and validity. It can be argued that moderate (.40-.60) correlations should not be interpreted in this way and that reliability coefficients <.70 should be considered as indicative of unreliability. Convergent validity coefficients in the .40 to .60 or .40 to .70 range should be considered as indications of validity problems, or as inconclusive at best. Studies on reliability and convergent should be designed in such a way that it is realistic to expect high reliability and validity coefficients. Multitrait multimethod approaches are preferred to study construct (convergent-divergent) validity.
Medical subject headings
- Physical Therapy Modalities
- Research Design