Construct validation of the Swedish Scholastic Aptitude Test by means of the Swedish Enlistment Battery

Abstract

The construct validity of the Swedish Scholastic Test (SweSAT), used for admissions to higher education, was studied through relating it to the Computerized Enlistment Battery (CAT-SEB) used at conscription to military service. Out of three male age cohorts tested at 18 years with the CAT-SEB in 1997, 1998, or 1999 six groups of men that had taken their first SweSAT in the Autumn (n = 3,163; 2,938; 2,820) the same year, or in the Spring (n = 3,238; 2,587; 2,447) the next year were used for the analyses. Four types of models were fitted to the tests: (1) a nested factor model with the latent variables general intelligence (g), crystallized intelligence (Gc) and general visualization; (2) a higher order model with g as an apex factor; (3) a multivariate regression model where the SweSAT subtests were regressed on the SEB factors; and (4) separate models of SweSAT and CAT-SEB with covariances between latent variables. The results show that the SweSAT is dominated by Gc factors and that the g factor has a smaller influence.