구성 타당성 평가방법에 관한 연구

A Study on the Methods for Assessing Construct Validity

The purpose of this study is to establish a basis for assessing construct validity of measures used in organizational research. The classic Campbell and Fiske's(1959) criteria are found to be lacking in their assumptions, diagnostic information, and power. The inherent confounding of measurement error with systematic trait and method effects is a severe limitation for a proper interpretation of convergent and discriminant validity. The confirmatory factor analysis(CFA) approach overcomes most of the limitations found in Campbell and Fiske's(1959) method. However, the CFA approach confounds random error with unique variance specific to a measure. The second-order confirmatory factor analysis(SOCFA) was shown to harbor rather restrictive assumptions and is unlikely to be met in practice. The first-order, multiple-informant, multiple-item(FOMIMI) model is a viable option, but it may also be of limited use because of the large number measures.