TAs, in a position to evaluate teachers but not under a grading bias, give ratings correlated with those of students.
PA and PASS systems of quiz management are essentially equal in effectiveness: Non-academic measures need consideration.
Lamberth and Kosteski (1981) state that the hlgh correlations found in their study between student and TA ratings "argue for the validity of student evaluations" (p. 11). I disagree with Lamberth and Kosteski's conclusion, not because the conclusion is based on faulty logic, but rather because the correlations they obtained appear to be artifactually high.For each Instructor In a large team-taught introductory psychology course, Lamberth and Kosteski obtained the mean ratlngs on a 28-item evaluation Instrument both for students taking the course and for TAs assigned to the course. Correlatlons between student and TA ratings were computed using the 28 pairs of means. Elght correlations, ranging from .44 to .86, were obtalned for six lnstructors, two of whom were rated twlce.Agreement between TAs and students assessed by such correlations may have as much to do wlth item and course characteristics as with instructor characterist~cs. For example, based on inspection of the flgures provlded by Lamberth and Kosteski, ~t appears that Instructors, as a group, were not highly rated on ltem 20 ("Treats class members as indivlduals"). Given the fact that class enrollments ranged from 200 to 700, such a flndlng is not unexpected. On the other hand, instructors, as a group, appeared to fare relatively well on ltem 4 ("Outlines directions of course adequately at beginnlng of semester"). One might speculate from the nature of the course (i.e., large enrollment, team teaching) that the same, apparently well-written, course outline was handed out at the beginn~ng of the semester for all Instructors.It is possible to check the above suggestlon. that high correlations reflect agreement about course and ltem characteristics rather than agreement about characteristics of lndivldual instructors, by using data pornts supplied by Lamberth and Kosteskl In their figures. For example. I obta~ned approxlmate TA ratlngs for Instructors 2. 3 and 4 for fall and approximate student ratlngs for instructors 1, 5, and 6 for sprlng semester by inspection of Figures 1 and 3. respectively, for each of the 28 questionnaire Items. ltem means were determined for each rating group and a correlation using the 28 pairs of resulting means was computed Even though the ratings were taken during different semesters, w~th different instructors, and rated by dlfferent raters (TAs vs students), the correlation obtalned was 53 (df = 26. p<.01). This correlation d~ffers significantly (p< 01) only from that obtained for Instructor 4 durlng the fall semester, uslng the test for the difference between two correlations described by McCall (1980, pp. 246-249). The other seven correlations were not slgnif~cantly dlfferent from 53 (p> 05) Hence, there appears to be no substantlal ~ncrement In agreement in ratings between students and TAs when lndlvldual lnstructors are considered beyond that which is obtained for lnstructors in the course taken as a group Glven stable estimates of Item parameters one way to obta~n correlat~ons that are sensitive only to perceived slmll...
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
customersupport@researchsolutions.com
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.