A Comparative Study of Class Performance in the Chemistry Laboratory Peter R. Adams The Pennsylvania State University-York
Campus, York, PA 17403
While a combination of student ahility and teaching skill probably make the primary contribution to student success in a traditional university setting, it is recognized that many other factors play a significant role?-3 Recently, an opportunity arose here t o allow suggestions and discussion of some of these factors based on statistically significant data that emerged from an analysis of student performance in two concurrently offered second-semester laboratory sections of the same course. These sections had the same course content of experiments performed in the same sequence, also the same instructor (the author), teaching methods, and grade assignment procedure. The pool of students entering both sections was the same, and placement in one section or the other was probably a t random based on schedule convenience of the particular student. The experiments performed in this course had been chosen previously to support material from the corresponding separate lecture course (taught by a different instructor) and not necessarily designed to be of progressive difficulty throughout the semester. While certain equipment was used by more thanonestudent insome cases, all students were required to measure and report on their own data individually. Each student's report on each experiment was graded as a percentage according to criteria of proper reporting of data and accuracy of results with somewhat more emphasis usually placed on the latter. In Section 1, two weak students from the initial enrollment (as identified by their earning low grades on the first experiments) quickly dropped the course and only one of the remaining students failed. By the same criteria, three students from Section 2 were identified as weak, but they elected to remain through the end of the course: two of them failed and one received a D grade. The grade percentages of all students who completed each section were analyzed separately in asimple scatter plot and also used for analysis of variance to compute F values based on the order of the experiments (to determine whether grades differed significantly between experiments), and in a second calculation to compute F values for linear it^.^,^ Letter grades assigned were not used in this analysis. As shown in the table, grade percentages differed significantly between experiments in both sections. Grade percentages showed a modest but steady improvement significant a t the 0.05 level, which was lacking from Section 2 even when those data were re-calculated excluding the three weak students. Further variance estimates of Section 1 grade percentage
correlated quite well with a positive-sloped linear relationship @ < 0.05). The differences between the two sections reported herein could not be due to differing sequences of experiments. I t could be due to differing abilities of the two groups of students; however, there was no structural procedure directing more able students t o Section 1, and this section did in fact initially enroll three weak students. In the lecture course complementary to these laboratory sections the average final grade of the Section 1students was significantly below that of the Section 2 students (1.59 and 2.45 on a 4-point scale, respectively) even when the calculation excluded the students identified as weak in this study (1.66 and 2.66, respectively). Examination of data from many prior offerings of this laboratory course in dual concurrent sections suggests that enrollment in both sections is a t random. Furthermore, this study has focused on whether improvement took place during the course rather than actual ability level. If the difference is due to subtle instructor biases, the author is not aware of them. This study was not planned or consciously conceived until the course had concluded and the data examined. Results from previous offerings of this course suggest that the student configurations of these particular sections were randomly produced. The instructor was not acquainted with or favorably disposed to the students of one section anv more than the other. The difference between the two sections most obvious to the author was the (voluntarv) self-removal of weak students from Section 1 which did not occur in Section 2. If other factors could be ruled out, these data might suggest that the presence of weak students in a class inhibits general improvement. The author acknowledges with thanks the consultation services of Dr. John Madden for statistical methods.
' Borch, G. D. Effective Teaching Methods; Merrill: Columbus, OH, 1988. ~
~
~
Gage, N. L. The ScientificBasis of the Art of Teaching; Teachers College, 1978. Jones. F. H. Positive Ciassrwm Instruction; McGraw-Hill: New York. 1987. Kirk, R. E. Expermental Design: Procedures for the Behavioral Sciences; BrookcCole, 1968. Winer. B. J. Statistical Prlnci~lesfor Ex~erimentalDesign, 2nd ed.; McGraw-Hill: New York. 197i.
Analvsls d Grades from a Chemlstrv Labwatorv Course Section NO.
1056
Initial Enmllment
Final Enrollmem
Journal of Chemical Education
Final Letter Grade Average
Lima, Regression Slope r
Analysis of Variance F-
6,nw