Principal goals at school: evaluating construct validity and response scaling format.

Abstract

The purpose of the present study was to test the efficacy and appropriateness of the 4-point response option of the Principal's Goals Scale of the SASS (1999-2000) survey. Competing dichotomous models with various conceptualizations were constructed and tested against the original polytomous conceptualization. Participants were 8,524 principals from whom 64% were males and 36% females. Principals' goals were assessed using a 6-item scale anchored across points reflecting proximity to achieving a goal. The original polytomous conceptualization was contrasted to a dichotomous two-pole conceptualization using a model with freely estimated discriminations (two-parameter logistic model, 2PL) as well as the Rasch model assuming equal discrimination parameters. Results indicated that the 2PL dichotomous model provided the most optimal model fit. Furthermore, item-related, and person-related estimates pointed to enhanced accuracy and validity for the dichotomous model conceptualization compared to the polytomous model. It is suggested that a dichotomous scaling system is considered in subsequent measurements of the scale as a means of enhancing the accuracy and validity of the measured trait.