1 Student Peer Evaluation

to respond to specific questions which they feel do not take into account their fundamental understanding of the mate- rial. Thus, there is presently ...
2 downloads 0 Views 1MB Size
Richard C. Parker and David S. Kristol New Jersey Institute of Technology

1

Students often feel that the grades they receive do not reflect their comprehension of the subject matter. Their most frequent complaint is that exams (which constitute the major component of their grades) do not fairly evaluate their knowledge because they are placed under great stress to respond to specific questions which they feel do not take into account their fundamental understanding of the material. Thus, there is presently great interest in new approaches to evaluation of students.' Because students gain a view of each others' abilities, that is unavailable to the instructor, by interacting with one another informally on a daily hasis, h e felt it would be informative to comDare the students' views of themselves with their course achievement, while recognizing that the students' o ~ i n i o n may s have been influenced by knowledge of each others' examination grades. Using- his own criteria, each student was asked to assign a number to every student in his class in proportion to his opinion of their abilities. Although an individual student might have been prejudiced in grading himself along with his competitors we assumed that the integrity of the students in general would statistically overwhelm the bias of any individual student. The students in two organic chemistry and two physical chemistry classes were instructed to assign a point value to each member of the class, ranging from one to n where n was the numher of students in the class; they were not permitted to consult with one another. They were asked to assign an average value to students with whom they were unfamiliar in order to prevent students from receiving a low score simply by not being well known. The instructor did not see the student peer evaluations until after the course grades were submitted, at which time the data were analyzed. The average rating for each student was calculated and compared with his course grade. Table 1 shows the average rating of those who received grades of A, B, C, and D in each course. The regularity of the distribution of averages with grades seems to indicate good correlation between student ratings and student achievement, in spite of the fact that the grade distrihution was not uniform. T o better compare the student evaluations and achievements we assigned the student with the highest student rating 100% and calculated the other student ratings relative to this student; these percentages were called S scores. At the same time we assigned the student with the highest course point score 100% and calculated the other student

'

Gowenlock, B. G., McIntosh, D. M., and Macksill, A. W., J. CHEM. EDUC., 50, 139 (1973); Kowell, S. T., and Marshall, T., Educntionol Research and Methods, 5,108 (1973).

Student Peer Evaluation

Distributionof Sand Tscares.

point scores relative to this student; these percentages were called T scores. Graphical plots were then made to see where the greatest deviations between student ratings and achievement occurred. The figure shows the distrihution of S and T scores for each class. The lines drawn represent the least squares fit. A slope of 1 would correspond to a perfect correlation. A slope of less than 1, which was ohserved in all four cases indicates that the range of student ratings was smaller than the range of course achievements. This was to be expected because the students assigned average ratings to students with whom they were not familiar. Points falling above the line indicate students whose ratings were relatively higher than their achievements when compared to the hest in each category. Points falling below the line represent students whose achievements were higher than their peer evaluations. The most extreme differences between S and T scores were examined closely. We found that those students whose S and T scores differed widely could be divided into two categories which correlated well with their personalities. Those whose S scores were much higher than their T

Volume 53. Number4 March 1976 / 177

Table 1. Comparison of Grader to Average Student Evaluations

--

A

14.5(3)a 14.0161 13.411) 13.616) 13.9(16)

Class I class I! Class 1 1 1 c1.31~ I v o v e r a l l Average -

B

D

C

10.8161 11.6141 11.1151 8.9121 10.8(171

8.518)

6:8131

9.7161

9.3(101

8.51101 8.9134)

5.1121

6.111) 6.1161

p~

a ~ u m b e r rtn parenthesis refer t o numbers o f students

Table 2.

Grade Achievement versus Difference in Ranking by Students and Teachers n "

C

C C C

C C

-6

-5

-

-

-

C C C

C C

C C C

C C C

A

A A

A

0

+ l +2 13 +4 +5

C

C

A

A

A

A

A

A A

4 -3

-2

A

-7

C C C

-1

178 / Journal of Chemical Education

C

C C

C C C

C C

16 +7

scores were people who had friendly, agreeable personalities, worked well with others, and were involved in extraclass activities. Those whose S scores were much lower than their T scores were people who did not mix well with others, were shy, and introverted. For the limited cases we have examined it appears that student peer evaluation is more influenced by personality than is the instructor's evaluation which is based upon objective criteria such as examinations. Another way to analyze the data is to rank the students from 1 (best) to 19 (poorest) based on both their student evaluations and their class point scores. Table 2 presents the difference between the ranking in each category emphasizing the grades achieved in the courses. A positive difference reflects a better ranking by the students than by the teacher. Most A and D students were ranked similarly by students and teachers, whereas there were widespread differences in ranking of the B and C students. This may he a reflection of the fact that i t is easier to pick out the very good and very poor students in any particular class than it is to distinguish among the average students. We have had the advantage of utilizing small classes and recognize the limitation of this approach to large classes. However, we hope to stimulate others to attempt studentpeer evaluations involving alternative approaches.