The Use of Classical and Rasch Latent Trait Models to Enhance the Validity of Affective Measures

Abstract

The purpose of this study was to compare the utility of the information yielded by the classical and latent trait psychometric models employed to examine the psychometric qualities of the School Situation Survey (SSS) (Helms and Gable, 1989), a 34-item attitude scale that assesses four sources and three manifestations of school-related stress. The first analysis employed classical principal component analyses followed by oblique rotations and was applied to data collected from a sample of 1111 students in Grades 5, 7, and 9. The second approach utilized a Rasch rating scale model and analyzed SSS data from 1958 students in Grades 6, 7, and 8. Practical aspects of instrument development including (a) lack of adequate variable definition and (b) the use of item-fit and person-fit statistics to reveal item content ambiguity were presented. In general, it was concluded that factor analytic procedures, while necessary, may not provide sufficient support for the adequacy of variable definition.