An analysis of discriminant validity of the teaching professional experience evaluation inventory for preservice teachers : a multitrait-multisource study
Abstract:
The purpose of this research was to develop and verify the quality of a teaching professional experience evaluation inventory for preservice teachers using discriminant validity for multitrait-multisource study, content validity, internal consistency reliability and inter-rater reliability. The participants consisted of 51 supervisors, 51 assistant classroom teachers and 51 preservice teachers. The instrument used for data collection was the teaching professional experience evaluation inventory. Data were analyzed by descriptive statistics, construct validity were analyzed by second order confirmatory factor analysis, reliability were analyzed by Cronbachs coefficient alpha and Two-way ANOVA and discriminant validity were analyzed by multitrait-multisource analysis. The results of this research found that 1) the teaching professional experience evaluation inventory (rated by supervisors, classroom assistant teachers and preservice teachers) consisted of 3 components; teaching professional practices, research skills in learner development and ethics of the teaching profession. 2) this inventory was considered to have content validity. The range of Index of Item-Objective Congruence was 0.75 1.00. The second order confirmatory factor analysis of this inventory were fit to the empirical data (x² = 8.78, p=.99). The factor loading of 3 components consist; teaching professional practices (0.97), research skills in learner development (0.95) and ethics of the teaching profession (0.87). The reliability of this inventory was 0.98. The inter-rater reliability showed no significant difference at the .05 level of the results of evaluation between 3 groups of raters. The results of this research showed convergent and discriminant validity of traits. This construct validity was confirmed by the difference of chi-square test for convergent validity models (∆x²₂-₁> 26.22) and discriminant validity models (∆x²₃-₁> 9.21), respectively at the .01 significant level.