Exploring impact of absent students on scale properties of student ratings of instruction in Turkey
In the present study comparability or ranking of instructors based on student ratings were investigated under the effect of absenteeism. To this end, invariance of scale properties of student ratings was examined via multigroup confirmatory analysis. Using randomly selected 2098 classes, equality of factorial structure, factor loadings, intercepts and residuals were tested. Results indicated that absent and regularly attending groups have developed the same conceptual meaning for the term instructional effectiveness. Also, ratings in the both groups of classes had a common unit, which makes within-class comparisons of instructors separately for attending and absent groups possible. However instructors who teach classes with absent students systematically receive lower ratings, indicating a bias between the two groups. Student ratings were adjusted against absenteeism to lessen the effect of bias. Results showed significant differences in the rankings of top-rated instructors both before and after the adjustment. Biased ratings pose a serious threat in comparability between instructors who teach absent and attending classes. Thus decisions involving instructors should be supported by other assessment mechanisms.