Comparison of Balanced and Unbalanced Designs Based on Generalizability Theory via the Data of Intramuscular Injection Station

Abstract

The purpose of this study is to compare the reliability and estimates of variance results by employing the balanced and unbalanced designs that were formed by rating the groups of participant students, both equal and different in numbers, considering their performance on the same task provided that the number of the participating students stays the same. The study group of the research are 240 first year students, who have taken the Intramuscular Injection Station in the Structured Objective Clinical Test of the Medicine Faculty of Hacettepe University. Eight raters was employed in assessing the performances of the students in the station. The variance values retrieved from the G studies done by each rater’s rating the students equal in number (balanced) and those unequal in number (unbalanced) were found to be parallel in both designs. Morover, considering the G and Phi coefficients obtained from the senarios in the balanced and unbalanced (s:r) x t designs were observed to be close.