Journal of the College of Community Physicians (Dec 2003)

An approach to validation of a multidimensional tool

  • N. S. Gunawardena,
  • R. de A. Seneviratne,
  • T. A. Atauda

DOI
https://doi.org/10.4038/jccpsl.v8i1.8302
Journal volume & issue
Vol. 8, no. 1
pp. 18 – 26

Abstract

Read online

IntroductionShort Form Health Survey-36 (SF-36) is a multidimensional measure of general health which has been used extensively in research on many population groups. ObjectiveTo validate SF-36 among lower limb amputee soldiers and a matched group of males, using triangulation. MethodologySF-36 was assessed among amputee soldiers (135) and matched healthy males (135) for judgmental, convergent-discriminant and construct validity. Judgmental validity assessed appropriateness of translation of conceptual definitions and cultural suitability. Multi-trait Multi-method Matrix technique assessed convergent-discriminant validity against another accepted measure of general health, the Nottingham Health Profile (NHP). Construct validity was assessed by checking whether: SF-36 distinguished expected differences between and within groups; and eight dimensions of SF-36 emerged from confirmatory factor analyses of data of populations studied. Cronbach’s alpha assessed reliability. ResultJudgmental validity was established. Correlations of similar dimensions of SF-36 and NHP were strong while dissimilar dimensions were weak confirming convergent-discriminant validity. Expected differences between and within groups were seen for scores of SF-36. Five factors among amputee soldiers and. six factors among the comparison group were derived from confirmatory factor analyses, which were similar to dimensions of SF-36. Cronbach's alpha for all dimensions exceeded 0.8 for both groups. ConclusionTriangulation proved that SF-36 was equally valid for both groups. This approach can be adopted to validate multi-dimensional instruments for cross cultural research where criterion validity cannot be assessed.

Keywords