The conundrum of class size is that despite a long history of generally consistent research results that class size has little if any demonstrated influence on college students' achievement, educators nonetheless continue to express concern about the effects of large class size on student performance, i.e., “most college professors believe small classes to be superior to larger ones in many respects” (McKeachie, 1980).
The fundamental finding of class size research literature, if the criterion of the class size effect is academic achievement, is clear. Unless class size is below 20, or even better, below 15, it is difficult to demonstrate that class size influences achievement, especially adversely as many professors assert. These class size levels of 20 or 15 parallel those of the popular but controversial Glass and Smith (1978) meta-analytic main finding, from mostly public school level studies but including some college level studies, that lower class size, i.e., < 15, is associated with higher achievement.
Laughlin (1976) characterized class size as a sacred cow and stated that college class size decisions are made on the basis of research, rhetoric and reputation. Readers can draw their own conclusions as to which of these three bases wields the greatest influence after contemplating the representative research reported herein.
The purpose of this article is to overview the college level literature on the effects of class size on student achievement. Recent studies will be emphasized. Class size is here defined as the number of students assigned to and enrolled in a specific class under the direction of a specific teacher.
The earliest reported professional interest in class size uncovered in this overview was associated with Comenius, who reportedly asserted that it was essential that one teacher teach several hundred scholars at a time because this would increase the teacher's interest in his work (Glass, 1985).
The first empirical research on class size was conducted in 1895 by Rice (Holland, 1954) and there has been considerable research especially over the last 70+ years. Consequently, class size research is nothing new. In a series of 59 college class experiments across many disciplines Edmonson and Mulder (1924) found differences favoring larger classes in 46 studies, although the differences were significant in only eight. A similar finding was reported by Hudelson (1928) who found higher subject matter achievement in larger lecture classes, across some 59 studies. Von Borgersrode (1941) reported, in the Encyclopedia of Educational Research, that, since Rice's initial work, 73 studies had been conducted, 13 at the college level, 10 of which were controlled. The range of students per class was 8 to 336, the typical upper limit of small classes being 25 and the typical upper limit of large class being 70. Bosley (1962) reported, without explicit citation, for the 1900-1932 period, that 205 books, monographs, and articles had been produced, and also that in another study published in 1958 it was indicated that 60 significant publications had been reported from 1932 to 1958. He concluded that “a majority of the small-class versus large-class investigations had not been overwhelmingly conclusive.” Vincent (1969) in his review for the fourth edition of the Encyclopedia of Educational Research reported that a dozen studies at the college level had been published since 1957 and that he had found only one in which achievement was significantly (p < .01) higher in small (N = 27) vs large (N = 85) classes. McKeachie (1963), a long-time authority on college teaching methods, concluded that “large lecture classes are not generally inferior to smaller lecture classes if one uses traditional achievement tests as a criterion.” When other objectives are measured, however, large classes tend to be inferior. Moreover, both students and faculty members feel that teaching is more effective in small classes. Laughlin (1976) concluded from research that “class size has not been shown to be a major factor in student learning,” and that “at this time research data does not substantiate that differing class sizes make a difference in student cognitive growth.” McKeachie (1980) concluded that, despite the fact that the relation between class size and achievement was probably the first problem associated with college teaching to be researched, it still remained an educational issue.
In 1985, Williams, Cook, Quinn and Jensen reviewed some of the relevant literature preparatory to conducting 24 stepwise multiple regression analyses between class size and achievement using final exams for eight ranges of class size across 15 different content areas, apparently at Brigham Young University. The 15 disciplines were: accounting; business management; child development and family relations; computer science; communications; economics; English; food science and nutrition; health; physical science; physics; religion; social science; statistics; and theater and cinematic arts. R2s were non-significantly small, close to zero. It was concluded that the results at the college level were compelling that “class size may be much less important an influence on student achievement than some educators have thought.” It was also concluded that class sizes less than 13-interesting since the Glass and Smith (1978) critical ceiling breakpoint was 15-were not included, and also that increasing class size from current levels of 30 to 40-not from 15-even “up to several hundreds may not radically affect college student achievement.”
In conjunction with their regression analysis, Williams et al. (1985) reported a limited review of the college level literature, the most recent review this author had found. They were parsimonious about making an overall conclusion concerning class size effects but inferentially implied that class size effects are still controversial, even after nearly a century of research. In McKeachie's (1990) latest look at class size, nested in a larger overview of research on college teaching, he concluded that, if the criteria of retention, problem solving, and attitude differentiation are employed, the evidence favors small classes. He also concluded, however, that the effect of class size on learning depends upon what the teacher does.
It is suggested that the reasons for lack of recent research on college class size effects, in addition to the lack of consensus on the definition of class size, large or small, include a veritable maze of confounded independent variables:
- Course Characteristics
- elective vs required
- major vs minor
- problem solving
- Instructor Characteristics
- Student Characteristics
- Methods Characteristics
- computer (CAI)
- cooperative learning
- critical thinking
- independent study
- problem solving
- programmed learning
- combinations of the above
- interactions of the above
There is also a plethora of dependent variables:
- course completion
- student ratings
- study time
- testing methods
- working conditions
The combination of some, many or all of these variables culminate in such a convoluted, tangled, inchoate mess that it is impossible to investigate systematically class size because not all, or even many, of these variables can be controlled in any single study. Thus, there seem to be basically only two research options, either little or no research the apparent current option, or else replicate the early approach and identify incidental sample classes of different sizes, arbitrarily define some as “large” and some as “small,” contrast the achievement of the “large” class size students with the achievement of the “small” class students on a rigorous measure, preferably a standardized achievement test, and then relate the differences, if any, in achievement to the differences in class size. Needless to say, the rigor of most of the class size research leaves much to be desired.
In the late 1950s a number of college level class size studies were conducted. Macomber and Siegel (1957) compared on student achievement a large television class with other large classes, sections taught by graduate students, and control classes. Of apparently 20 comparisons, 19 did not differ significantly. Rohrer (1957) compared one large class with one small class for each of three instructors in an American Government course. There was a slight, non-significant difference in achievement factor in favor of the larger sections. Another study, with greater methodological control, was reported by Cammarosano and Santopolo (1958) who equated their research classes on student high school GPA and scholastic aptitude. The test scores for each pair of large (N = 60) and small (N = 30) classes for economics, politics and sociology were non-significantly different on eight of the nine comparisons. Nelson (1959) conducted a similar type of analysis. He compared the achievement of elementary economics students whom he had attempted to match on school, level and sex, in large vs small sections, for four instructors. For the seven comparisons each of the large classes had over 84 students while each of the small classes had under 21 students. Differences on achievement between section sizes were small and Nelson concluded that “large classes of from 85 to 140 can be taught as effectively as small classes of 18 to 20.”
In one of the few relatively modern class size experiments, De Cecco (1964) actually randomly assigned students (first year psychology) to one of three kinds of instructional groups: two large-experimental, one with 127 students and one with 97; six small-experimental, ranging in size from 34 to 18 with a mean of 28; and ten small control groups, ranging in size from 35 to 22 with a mean of 28. Two achievement criteria were used. Class size was not influential on either criterion, students' preferences or course evaluations. More recently, Fischer and Grant (1983) using the Florida Taxonomy of Cognitive Behavior, as a part of a larger study of students' cognitive processes, compared different class sizes of students; opportunity to apply the cognitive skills of analysis, synthesis and evaluation. Students in small classes of 15 or fewer did engage in greater use of the higher order thinking processes. Ancillary but perhaps more important findings were that talk in college classrooms seldom encouraged higher order thinking, and also that most discourse was conducted at the lowest cognitive level. This was true regardless of: kind of institution, public or private; course level; subject level; or length of time the class had been in session. In the most recent college research located, an English-as-a-Second-Language (ESL) study, Bolton (1988) compared 50 students from a large ESL megasection (three sections in one) with 50 conventional class students. Mixed results were obtained over a variety of criteria: controls earned slightly more cumulative credit hours over the year; megasection students had a higher cumulative GPA; there was little difference in retention rates in course sequence; and megasection students did as well as controls one year later.
A peripheral area of the class size research literature is the influence of class size on students' ratings of the effectiveness of their teachers. Fortunately, Professor Kenneth Feldman addressed this issue in a meta-analysis as recently as 1984. He obtained an average, a non-significant correlation of -.09 between class size and students' overall ratings of college teaching effectiveness. There were also slightly higher inverse correlations between class size and specific individual instructional dimensions pertaining to instructor-student interactions.
Following are the conclusions about the college level class size literature from this overview:
- the size of the literature is large but most of the literature is old;
- recent literature is limited;
- methodologically sound research is extremely limited;
- experimental literature is even more extremely limited;
- most college professors favor small classes;
- large class size probably does not maximize student performance if student higher order thinking is the criterion, although both the actual use of this criterion and also its related research are limited;
- class size has negligible effects on student performance if student achievement is the criterion, unless the class size is below 20, or even better, below 15.
Finally, it may be tuitional to borrow here, the college level, Slavin's (1989) conclusion from his analysis of the meta-analyses of the effects of class size on student achievement across all levels of education. He concluded that “not until class size approaches one is there evidence of meaningful effects.”
Bolton, J. K. (1988). Larger is sometimes better: Approaches to large classes. ED 292 359.
Bosley, H. E. (1962). Class sizes and faculty-student ratios in American colleges. The Educational Record, 43, 148-153.
Cammarosano, J. R., and Santopolo, F. A. (1958). Teaching efficiency and class size. School and Society, 86, 338-341.
De Cecco, J. P. (1964). Class size and co-ordinated instruction. British Journal of Educational Psychology, 34, 65-74.
Edmonson, J. B., and Mulder, F. J. (1924) Size of class as a factor in university instruction. Journal of Educational Research, 9, 112.
Fischer, C. G., and Grant, G. E. (1983). Intellectual levels in college classrooms. In Studies of College Teaching, C. L. Ellner and C. P. Barnes, editors. Lexington, MA: D. C. Heath and Company.
Glass, G. V. (1985). Class size. In The International Encyclopedia of Educational Research and Studies. T. Husen and T. N. Postlethwaite, editors. New York: Pergamon Press.
Glass, G. V., and Smith, M. L. (1978). Meta-analysis of Research on the Relationship of Class Size and Achievement. San Francisco: Far West Laboratory for Educational Research and Development.
Laughlin, J. S. (1976). A Sacred cow - class size. College and University, 51, 33-347.
Macomber, F. G., and Siegel, (1957). L., A study in large-group teaching procedures. The Educational Record, 38, 220-229.
McKeachie, W. J. (1963). Research on teaching at the college and university. In Handbook of Research on Teaching. N. L. Gage, editor. Chicago: Rand McNally and Company.
McKeachie, W. J. (1980). Class size, large classes and multiple sections. Academe, 66, 24-27.
McKeachie, W.J. (1990). Research on college teaching: The histroical background. Journal of Educational Psychology, 82, 189-200.
Nelson, W. B. (1959) An experiment with class size in the teaching of elementary economics. The Educational Record, 40, 330-341.
Rohrer, J. H. (1957), Large and small sections in college classes. Journal of Higher Education, 28, 275-280.
Slavin, R. E. (1989). Class size and student achievement: Small effects of small classes. Educational Psychologist, 24, 99-110.
Vincent, W. S. (1969). Class size. In Encyclopedia of Educational Research, Fourth Edition. R. L. Ebel, editor. New York: The Macmillan Company.
von Borgersrode, F. (1941). Class size. In Encyclopedia of Educational Research. W. S. Monroe, editor. New York: The Macmillan Company.
Williams, D. D., Cook, P. F., Quinn, B., and Jensen, R. P. (1985). University class size: Is smaller better? Research in Higher Education, 23, 307-318.
John Follman is with the College of Education at the University of South Florida in Tampa, Florida, U.S.A.