An analysis (between and within classrooms) of student evaluation of instruction
Much of the previous research on student ratings of their teachers has used inappropriate methods of analysis. These analyses fail to recognize that the score a student gives his instructor may depend upon characteristics of that student as well as upon the characteristics and behaviors of the instructor;The data of this study consisted of 2,107 students enrolled in several sections of 30 courses, resulting in 88 classrooms. The methodology used in this study analyzes student variability within classrooms separately from one classroom to another and incorporates the fact that results from these two kinds of analysis are expected to be the same if teacher and students do not differ from one classroom to another. It also recognizes that when student characteristics are related to how those students rate their instructors, and classrooms have different kinds of students in them, relative to these characteristics, then the ratings received by instructors are not comparable because student differences are confounded with teacher differences;Factor analysis of the between correlations results in 10 factors. These same factors tended to occur in the within analysis, but the within classroom variability was substantially smaller than the between classroom variability, indicating differences in the ratings received by different instructors. Within classrooms, females tended to rate their instructors slightly higher than did males (r (TURNEQ) .05), but this correlation was substantially larger (r (TURNEQ) .4) for the between analysis suggesting that courses preferred by females (e.g., home economics courses) are rated higher than courses preferred by males (e.g., engineering courses). Also, those classrooms with students who anticipated high grades, and who took the course as an elective tended to be rated higher than classrooms with students with other characteristics. These relationships were smaller or nonexistent within classrooms. Self-reported GPA was negatively related to the student rating of the textbook for the between analysis, and essentially unrelated to the other nine variables; otherwise, for these eighty classrooms which vary in both content and level, it appears that student characteristics play a substantial role in determining the average rating and suggests that ratings of teachers are comparable only to the extent that the students who rate those teachers have similar characteristics.