Student Performance In Upper-Division Business Core Courses: Using Control Variables To Determine The Effect Of Class Size

Donna A Driscoll,Janna B Arney,Earl J Weiss,Paul J Lazarony

doi:10.19030/jber.v3i1.2731

Abstract

Although the effect of class size upon student performance has been the focus of numerous studies, the results have been extremely mixed, including positive effects, no effects, and negative effects.&nbsp; The authors of this study believe that this lack of consensus could be due, at least in part, to the shortage of control variables employed in previous studies. &nbsp;The purpose of this study was to determine whether the addition of four control variables (course being taught, length of class period, instructor, and student ability level) to the analysis would be helpful in isolating the effect, if any, of class size upon student performance. &nbsp;To more fully explore the effects of including control variables upon the results, the control variables would be added sequentially, one at a time.&nbsp; The analysis would thus consist of a series of five regression sets.&nbsp; The first set would be a single, simple regression of student performance on class size.&nbsp; The second and subsequent sets would each consist of a group of simple regressions that controlled for one, two, three, and then finally all four of the control variables.&nbsp; The data to be used would be 113,468 course grade records (for the years 1990-2002) for 10 upper-division business core courses. Student performance would be defined as course grade; class size would be defined as either large (75 or more students) or small (60 or fewer students).&nbsp; The results showed that an initial slightly positive effect (.06 grade points) of class size on student performance (found in the Regression Set 1 before the addition of any control variables) changed substantially to 4 positive effects, 57 no effects, and 21 negative effects (after the addition of all four control variables in Regression Set 5).&nbsp; The Adjusted R-Squares also increased (from a rounded value of .000 in Regression Set 1 to a high of .428 in Regression Set 5) as did the grade point differences in performance between small and large classes (from .06 grade points in Regression Set 1 to a range of from .68 to -1.24 grade points in Regression Set 5).&nbsp; Although more research is certainly needed in this area before definitive conclusions can be reached about the effect of class size upon student performance, it appears that the control variable approach used in this study does shed some light on the seemingly inconsistent results of previous studies.&nbsp; Specifically, our results show that: (1) it quite likely (57/82 or 70%) that there will be no significant effect of class size upon student performance, (2) if there is an effect, it will most likely (21/25 or 84%) be a negative one, and (3) the effect, if any, is highly dependent upon the specific combination of course, class length, instructor, and student ability level involved.

Full Text