MODELS AND PERSONALITY-TRAITS IN OCCUPATIONAL PSYCHOLOGICAL-TESTING

From a study of methods of test construction - criterion-keying, item analytic, factor analytic and using item characteristic curves - it is shown that, in the sphere of personality tests, only factor analytic methods can ensure univariate measures. A further scrutiny of factor analytic methodology reveals various technical defects which have created a chaos of different findings and factors in personality inventories. It is finally demonstrated that rotation to simple structure yields reliable and replicable factors and that when this is done, four or five factors account for most of the variance in personality inventories. It is concluded that in occupational testing only tests measuring these factors, or primary factors clearly related to them, should be used.