Does Class Size Matter?
By: David Bressoud @dbressoud
David Bressoud is DeWitt Wallace Professor Emeritus at Macalester College and Director of the Conference Board of the Mathematical Sciences
I am often asked by math department chairs what evidence they can take to their deans to argue against increasing the size of their classes. This is particularly pertinent now when budgets are severely squeezed. Larger classes cost less per student, but administrators also worry about student success and retention. Is there research that establishes the effect of class size on student success?
This is an old question. It was the subject of a paper by Edmondson and Mulder that appeared in 1924, apparently the first published research on teaching and learning in higher education. Unfortunately, although there have been many studies in the decades since then, there is little that is definitive. There are studies that give the advantage to small classes, studies that indicate that students in large classes do better, and studies that show that it makes no difference, all of which are problematic to one extent or another.
One issue is that it is difficult to measure what is meant by “doing better.” The easiest and most common measure in published studies into the effect of class size only considers grades. My own experience from the MAA calculus studies is that grade distributions are remarkably uniform regardless of class size. There seems to be a common expectation in calculus that about 20 to 25 per cent of a class should receive an A with a similar percentage receiving a B, irrespective of how the course is taught or what the students have learned.
In addition, expectations for what students have learned and the means of assessing them change as class size increases. Large enrollments affect how homework is handled, the kinds of questions asked on exams, and the ability to engage students in activities that explore the depth of their understanding. What can be done in a class of at most 40 students is difficult to pull off once the enrollment exceeds 60. When enrollments are in the hundreds, assessments are severely constrained.
What is unquestionable is that class size affects possible teaching strategies. As McKeachie reported in his landmark summary of research on this issue,
Large lectures are not generally inferior to smaller lecture classes when traditional achievement tests are used as a criterion. When other objectives are measured, large lectures are on shakier ground. Goals of higher level thinking, application, motivation, and attitudinal change are most likely to be achieved in small classes. Moreover, both students and faculty members feel that teaching is more effective in small classes. (McKeachie, 1980, p. 26)
Image of Elizabeth Arnold at James Madison University taken from MAA Focus, June/July 2019.
This is consistent with the findings of Seymour and Hunter in Talking About Leaving Revisited. They found that large classes discourage student persistence and enable disruptive student behavior: "overt disattention to the lecturer, arriving late, leaving early, eating, drinking, sleeping, chatting to friends, and use of cell phones" (pp 213–214).
The most recent article on the effect of class size was published this past fall in the AERA Educational Researcher (Ake-Little, von der Embse, and Dawson, 2020). The study was large, 32,766 students across 8,049 classes over 14 terms at Temple University, a good example of a public R1 university in an urban situation. They looked at a wide variety of factors that might interact with class size to influence student grades including gender, race/ethnicity, discipline, instructor rank and experience, and student reported variables that included expected grade, level of interest, number of hours spent studying, and high school preparation.
The article reveals a number of insights that go beyond class size. Not surprisingly, the strongest predictor of a good grade was expecting to receive a good grade. But the second strongest was high school GPA, stronger than SAT/ACT scores. As I have observed from other studies, spending more time studying correlates with lower grades. This may be because the better prepared students do not need to study as much or because they know how to study more effectively.
Also of interest, though not unexpected, is that tenured and tenure track faculty tend to give lower grades than contingent faculty (significant at p < 0.001), and instructors who have been teaching for at least ten years give lower grades than those who have been teaching less than five (again p < 0.001).
Looking just at the effect of class size and adjusting for all of these factors, the authors looked at STEM classes by gender and race/ethnicity. For underrepresented women in STEM (n = 2739), there was a small drop-off in average grades as class size increased, but it did not meet a significance level of p < 0.05. For White and Asian women (n = 9433), classes of over 60 did depress grades, now with p < 0.01. For White and Asian men (n = 9411), there was a very clear drop off in grades as class size increased. Classes of less than 26 raised the average grade, while classes of 41 to 60 lowered it, both deviations significant at p < 0.001. However, it should be noted that even though the changes were statistically significant, the amount that average grades changed was very small.
The most intriguing result was for underrepresented men in STEM courses (n = 1591). There, small classes (< 26) lowered grades (significant at p < 0.001), while classes above 60 raised the average grade. I am certain that there are a number of possible explanations for the discrepancy in the effect of class size on White and Asian students as opposed to Black and Hispanic men, but the one that comes immediately to mind and is quite disturbing is that this reflects implicit bias. Students in large classes are relatively anonymous. Implicit favoring of White and Asian men is most likely to occur in small classes where gender and race are more easily identified and taken into consideration, even if unconsciously.
It would be good to have a study that looks at the effect of class size on downstream performance and ultimate success. Nevertheless, our lack of detailed knowledge does not change the fact that the focus in every course should be on the goals for that course and what we expect of successful students. It is this that should inform decisions on class size. Furthermore, the Ake-Little et al. study serves as a warning that in small classes we need to be particularly alert to implicit biases that may affect student performance or our assessment of that performance.
References
Ake-Little, E., von der Embse, N., and Dawson, D. (2020). Does Class Size Matter in the University Setting? Educational Researcher. 49 (8), 595–605. https://journals.sagepub.com/doi/abs/10.3102/0013189X20933836
Edmondson, J. B. and Mulder, F. J. (1924). Size of Class as a Factor in University Instruction, Journal of Educational Research, 9
McKeachie, W. J. (1980). Class Size, Large Classes, and Multiple Sections. Academe, 66 (1), 24–27.
Seymour, E. and Hunter, E.–B., eds. (2019). Talking about Leaving Revisited. Cham, Switzerland: Springer. https://www.springer.com/gp/book/9783030253035
Download the list of all past Launchings columns, dating back to 2005, with links to each column.