TPC-Journal-V2-Issue1

84 The Professional Counselor \Volume 2, Issue 1 they filled out the SET and estimated their grade, because only two of the 28 activities accounting for 125 of 653 total points were still ungraded at that point. Student Evaluation of Teaching Ratings We used a standardized instrument for SETs, the Student Instructional Rating System (SIRS; Arreola, 1973), a student course on form developed at Michigan State University (Davis, 1969) and adapted for use at our university. SIRS provided an opportunity for instructors to obtain reactions to their instructional effectiveness and course organization and to compare these results to those of similar courses offered within the university. The SIRS consisted of 32 items and 25 of these items enabled students to express their degree of satisfaction with the quality of instruction provided in the course by using a 5-point Likert scale. For example, the course was well organized could be marked strongly agree, agree, neutral, disagree, or strongly disagree. One item on the SIRS was of special interest in this study: What grade do you expect to receive in this course? A, B, C, D, or F. We also employed a second instructional rating instrument, the State University System Student Assessment of Instruction (SUSSAI) which had been used at the university for five years prior to this study. This instrument consisted of eight items focused on class and instructor evaluation. One item was of special interest in this study: Overall assessment of instructor: Excellent=4, Very Good=3, Good=2, Fair=1, Poor=0. Data Collection After obtaining permission from the university institutional review board, we received the archived career course grade data for a six-year period. We aggregated the grades of these 1,479 students by class schedule and averaged the results to achieve a mean EG for each class schedule format. The data relating to students’ perceptions of what they had achieved and the quality of instruction they had received was collected as follows: On the last week of class, while filling out their teacher evaluations, all students in a section were asked to indicate the grade they expected to receive and the results were tallied and averaged to determine a class mean XG. These class averages of 57 sections were forwarded to the researchers, and the results were tallied and averaged to find the mean XG for each class schedule format. In addition, we retrieved overall class ratings of instructors for an ad hoc sample of career classes over the 6-year period. These data enabled us to examine the relationships between mean EG and XG, EG and SET, and XG and SET. Procedures In this team-taught course where all instructors were involved in making large- and small-group presentations, each co-instructor had primary responsibility for evaluating the progress of students in his or her small group and assigning a grade, while the lead instructor of the team had overall responsibility for course presentations and management. In completing the SIRS and SUSSAI items for the SET, students were asked to provide a composite rating of the instructional team for their section. SETs were completed anonymously during the final two class meetings while instructors were out of the room and then returned by a student proctor to the university’s office of evaluation services. Data Analysis We examined how different class formats influenced mean EG, XG, and SET. The independent variable of class schedule format had four levels. The first three levels met over the course of a 16-week fall or spring semester for either 3 hours once a week (W), 1.5 hours twice a week (MW/TuTh), or 1 hour three times a week (MWF). The final level met for 2 hours four times a week over the course of a 6-week semester (MTuWTh). Because the assumptions related to independence for the three evaluative measures could not be met (i.e., the evaluations for each class section were correlated), we analyzed the data using a split-plot design. Results As is the case for other ANOVA and MANOVA tests, the dependent variables were assumed to be normally distributed. We tested the dependent variables to determine if they were normally distributed by computing skewness and kurtosis of