Dependability of preservice clinical teaching scores using Generalizability Theory

Introduction The core of teacher training is to prepare preservice teachers to engage in teaching practices that enhance the acquisition of 21st-century skills. Clinical teaching is an integral part of teacher training. The processes involved in assessing preservice teachers’ clinical teaching experiences are not immune to error; therefore, when not well executed, they may compromise the validity and reliability of the scores derived. Methods Our study applies Generalizability theory (G-theory) to examine sources of error in clinical teaching practice at a university in Ghana. We adopted a three-facet partially nested random balanced design, with occasion nested within rater, nested within person, and all crossed with items, denoted by ( o : r : p ) × i , to investigate clinical teaching observations involving 20 items, 107 raters, 2 occasions, and 208 preservice teachers. Results We found that person-by-item interaction and the residual were the major sources of error. The generalizability and dependability coefficients were 0.91 and 0.90, respectively. Conclusion We concluded that preservice teachers’ teaching competencies were assessed with high level of precision, though the assessment process was characterized by some errors. The study made useful recommendations to key stakeholders.