Equivalence of constructs measured by job-specific and commercially available aptitude tests.

Abstract

This study examined the equivalence of constructs underlying scores on tests designed to measure the knowledge, skills, and abilities required by specific jobs with the constructs underlying scores on aptitude tests taken from published test batteries. Several models of construct equivalence, differing in their assumptions about factor patterns, factor loadings, and variable uniquenesses, were assessed with confirmatory factor analysis. Results indicated that the job-specific tests measured constructs that were essentially equivalent with the constructs measured by the commercially available tests, although the magnitude of unique residual variances differed among the two sets of tests. Furthermore, multiple-groups confirmatory factor analysis indicated that tests loaded equivalently on shared constructs across several sex and race subgroups, although unique residual variances differed across groups. Practical and theoretical implications are discussed. Choices of predictor measures in personnel selection contexts typically require decisions regarding whether tests of general aptitudes relevant to success across diverse jobs, or tests designed to measure the specific knowledge, skills, and abilities (KSAs) needed to perform the jobs of interest, should be used. Tests of the former type are commercially available and may be cost effective in the short term. Moreover, substantial meta-analytic evidence suggests that general aptitude tests predict success across a wide variety of jobs (Ghiselli, 1966,1973; Hunter & Hunter, 1984). General aptitude tests can be criticized, however, for containing items that have no apparent relevance to success for any particular job (Kleiman & Faley, 1985). Determination of the desired predictors for a job requires that the user base his or her decisions on some logical, empirical, or theoretical foundation (Society for Industrial and Organizational Psychology, 1987). Hence, general aptitude tests may come under particularly close scrutiny if a rationale for their choice has not been made explicit (Kleiman & Faley, 1985). Defending the use