Dependability of Students‘ Internship Mentors‘ Results Using Generalizability Theory at University of Education, Winneba

Abstract

The main purpose of this study was to determine the dependability of the mentors‘ results of the University of Education, Winneba, School Internship Programme (UEW-SIP), using Generalizability (G) theory. Inherent in this purpose were to find the reliability and the sources of error in the mentors‘ results. A random effect one-facet crossed design in which intern (p) was crossed with occasion (o), was used for the study. The study used eight purposively selected Faculties out of a total of 14. A total of 9,082 bachelor‘s degree graduates results for the academic years 2015/2016, 2016/2017 and 2017/2018 were used for the analysis. Data were analysed by performing a univariate generalizability analysis using EduG version 6.1. It was found that for relative interpretation, the results were strongly reliable (Coef_G relative of 0.66 – 0.84) and for absolute interpretation, the results were moderately to strongly dependable (Coef_G absolute of 0.59 – 0.81). The major source of error was the intern by occasion (p x o) interaction. The Intern Teaching Evaluation Form (ITEF), which is the rating scale for evaluating teaching practice in UEW, was found to be reliable to a greater extent. For most dependable mentors‘ results while ensuring economy of use of resources in the UEW-SIP, the optimum number of occasions was found to be five. It was recommended among other things that, the university supervisors‘ rating should be increased from one to at least two occasions, so that G theory can be applied to find the psychometric properties of the results. Again, the number of occasions for mentors‘ rating in the UEW-SIP should be increased from three to five to ensure most stable and dependable results.