A Confirmatory Factor Analysis Approach to Construct Validation

Abstract

A multitrait-multimethod matrix containing two methods of measuring 12 personality traits was analyzed. Because earlier studies disagreed on the effects of method variance on construct validity, confirmatory factor analysis was applied to the data. Although unexplained variance remained, method factors and a general personality factor significantly improved the fit of a model containing only trait factors. It was concluded that confirmatory factor analysis merits wider use in situations where researchers can formulate specific hypotheses about the structure of data.