Class Size and Achievement in Higher Education: A Summary of Current Research
Toth, Linda S., Montagna, Linda G., College Student Journal
With rapidly increasing enrollment and soaring tuition becoming major concerns of students, instructors, and administrators in many institutions of higher learning, the search to identify the most effective ways of delivering a high quality education at a reasonable cost continues. The decrease in government funding affecting many state-owned colleges and universities nationwide has often resulted in a decrease of course sections offered and an increase in the class size of the remaining available classes. The purpose of this review is to examine the current research regarding the relationship of class size to student achievement in higher education. The eight studies examined, published between 1990 and 2000, were analyzed as to methodological strength, measurement of achievement, significant findings, and resulting recommendations. The studies yielded mixed results, and upon close examination, it was found that the design methodologies were weak overall, with four of the eight studies examined defining "achievement" as simply the students' class grade alone. It is suggested that future research in this area investigate the possibility that other variables such as potential grade inflation, lower academic standards, student aptitude and readiness for college work, lack of remediation for ill-prepared and disadvantaged students, student learning styles, instructors' teaching styles, and student motivation and effort could confound research results in this area and may account for the inconsistent results.
There is a growing concern about the preparedness of college and university graduates upon entering the workforce. After the 1983 announcement that the U.S. was at risk due to poor educational performance, billions of dollars were allocated to the educational system. Despite this effort, scant results were yielded, undermining public confidence in the instructional ability of U.S. educators. However, it has been contended that there are other factors involved, with one such factor currently under examination in both the political and academic arenas being the impact of class size on student achievement in higher education. (Van Allen, 1990.)
In light of rapidly increasing enrollment in many colleges and universities across the nation, administrators are under fire concerning the issue of growing class size and the potential diminishment of academic standards. Van Allen (1990) asserts that the "quantitative product"--monetary gains afforded by increased enrollment--are outweighing the "qualitative product"--well-educated and knowledgeable college graduates. It is, therefore, of great importance that research be conducted to provide convincing evidence as to whether or not students, faculty, staff, and perhaps the nation at large may be suffering negative consequences due to the increase in class size.
In assessing the possible effect of class size on achievement, the issue of a good measure of academic achievement has been called into question. Kennedy and Siegfried (1997) state that the influence of class size on achievement depends upon the measure of achievement and that results of studies surveyed conclude that when measures of knowledge are used, the large class method is as effective as the small-class methods. However, when measures of transfer of knowledge to new situations, retention of information, problem solving, critical thinking, and attitude change or motivation are used, small-class discussion methods are favored (Kennedy & Siegfried, 1997). Taking this into consideration, it is apparent that using inappropriate or oversimplified methods of assessing achievement may lead to invalid inferences. Further, it has been suggested that the grading systems in higher education, and education in general, have been plagued by grade inflation (Gibbs, Lucas, & Simonite, 1996; Noble, 2000; Van Allen, 1990), resulting in a ceiling effect, with grades within a class being more heavily distributed in the "A" or "B" ranges than in lower ranges. If so, then studies relying upon class grades as the sole measure of achievement may be invalid due to the resulting restriction of the data range.
This summary focuses upon studies published between 1990 and 2000 that address the issue of class size and achievement in higher education. Unfortunately, current literature addressing this subject was scarce. The eight studies in this review were examined in terms of (a) general methodological characteristics, (b) the measurement of achievement itself as defined in the simplistic form of course grade alone, as well as in other ways such as a pretest--posttest strategy, and (c) research results. Finally, suggestions regarding directions for future research are offered.
Of the eight studies examined in this review, six were quantitative and two were qualitative. All of the studies employed a non-experimental design. The class subjects of the studies spanned several academic disciplines with three studies being in economics, one in statistics, one in electrical and computer engineering, and three that combined more than one discipline. Of the eight studies, seven defined class size in measurable terms. Furthermore, pertinent variables, such as sample size and sample characteristics, were loosely defined in these studies, and in some instances were not addressed at all. Of the six quantitative studies examined, four defined "achievement" or "performance" in terms of class grade alone.
In the first of these studies, conducted by Gibbs, Lucas, and Simonite (1996), college "modules" or classes across disciplines were analyzed for a ten-year span from 1984-85 to 1993-94. Data collection included (a) the subject area, (b) stage (1 = year one of full-time study, or 2 = years two or three), (c) year, (d) "module enrollment" or class size, and (e) the number and proportion of students getting the grades "A," "B+," "B," "C," and "F." Class size was defined as "small" if containing 30 or fewer students and "large" if containing more than 70 students. Achievement was defined as the average of module marks, expressed in percentage, for each module, and data were collected on 6075 students over the ten-year period. It was hypothesized that a negative relationship between module enrollment and the average grades for the module exists, thus predicting that modules with larger enrollments will have lower average grades. It was further hypothesized that over time, as module enrollments increase, average student performance will decline (Gibbs et al., 1996).
The results of a Linear Regression Analysis confirmed the first hypothesis, but contrary to predictions, failed to confirm the second hypothesis (Gibbs, Lucas, & Simonite, 1996). Thus, in light of the contradictory results, these findings are not very convincing.
A second study, conducted by Raimondo, Esposito, and Gershenberg (1990) at the University of Massachusetts, examined the relationship between class size in introductory economics courses and student performance in subsequent intermediate economics theory courses. One hundred forty six students who had taken both courses comprised the research sample which consisted of 32.9% women, 11.6% Asians, 9.6% Blacks, 3.4% Hispanics, 89.7% full-time students, and 52.7% economics majors. A binary coding system was used to identify students who had taken the introductory courses in a large (200-350 students) or small (25-35 students) class. Of the 146 students in this sample, 58.9% had taken the introductory course in a large lecture format. Achievement in this study was defined as the students' course grades (with "A" = 4.0, "A-" = 3.75, "B+" = 3.25, "B" =3.0, etc.). The mean grade of the large introductory economics course was 2.51 with a SD of 1.08 and the mean of the small economics course was 2.66 with a SD of 1.00.
Once again, the results of a simple linear regression model yielded mixed results. Enrollment in a large lecture introductory course did not significantly influence student performance in the intermediate microeconomics course, but it did have a negative and statistically significant effect on student performance in the intermediate macroeconomics course. One explanation for the conflicting results was that different levels of cognitive skills were required in the two intermediate theory courses, with instructors in the intermediate macroeconomics course pushing students beyond the recall and recognition skills which were stressed by the instructor in the intermediate microeconomics course. (Raimondo, Esposito, and Gershenberg, 1990.)
In 1999, Borden and Burton examined the impact of class size on student performance in introductory college courses across disciplines. During the initial analyses, cutoff points for small, medium, or large class sizes were determined using the CHAID clustering procedure included within the AnswerTree analytic software (SPSS). CHAID is a hierarchical divisive clustering method that uses binary spits to divide a sample into successive subgroups based on selecting a predictor variable that maximizes reduction in the unexplained variation of a criterion variable. Designed for nominal or ordinal data, CHAID includes a discretizing function that divides the quantitative predictor into intervals and from then on the data are treated as ordinal categories. Utilizing this method, class size was defined as (a) a "small" class consisting of 5 to 30 students, (b) a "medium" class consisting of 31 to 90 students, or (c) a "large" class consisting of 90+ students. Achievement was defined in two parts including (a) student course grade and (b) whether the student completed the course "successfully," meaning with a "C" grade or better.
The final analysis compared students in large and small sections by examining (a) the likelihood of enrolling in a subsequent course within the discipline and (b) performance in the next course while controlling for the grade in the initial course. Results of an ANCOVA revealed an overall negative impact, with students in larger class sections not performing as well as students in smaller sections of the same course. There was also an interaction between the students' prior level of preparation and section size, or more specifically, the negative impact of large section size was significantly greater among lower ability students. (Borden & Burton, 1999).
The effects of class size on college student achievement in statistics were examined by Hancock (1996) in a study in which an instructor of sophomore statistics taught one section each semester as an auditorium class with triple the normal enrollment cap or 120 students. Class size was defined as (a) "normal" sections averaging 39 students and (b) "megasections" averaging 118 students. Achievement was based on regular semester tests, and was measured by grade distributions (90% = "A," 80% = "B," etc.). After three semesters, the achievement of the students in the "megasections" was compared to six normal-sized sections taught by the same instructor using the same text book, tests, and grading procedures.
A Chi Sqaure Test for Independence yielded a test statistic that was clearly not significant, and it was concluded that the grade distribution in this statistics class was not affected by class size. (Hancock, 1996.) A strength of this study was that the researcher controlled for instructor, test, text, and grading procedure. However, weaknesses included the small sample size, the narrow definition of achievement, and the failure to specify sample characteristics.
A pretest-posttest strategy was used by Kennedy and Siegfried (1997) to examine the influence of class size on achievement in economics while holding constant a variety of factors thought to influence learning such as student ability and study hours. This study utilized a national economic education database (TUCE III), consolidating these data so that each observation represents a class rather than a single student, thereby reducing estimation problems. Classes were chosen as units of observation because (a) using student observations would give larger class sizes a heavier weight in the analysis, and (b) many of the explanatory variables such as SAT scores measuring ability and self-reported hours spent studying contain substantial measurement errors, whereas averaging has the benefit of reducing the influence of measurement errors.
Class size was defined as the average enrollment at the beginning and end of the course. Sixty-nine classes ranging in size from 14 to 109 students were used, plus one class of 278. Achievement was primarily measured with POST, which represents the score on a 30-question posttest, and DIFF, which represents the difference between the 30-question posttest and pretest scores (Kennedy & Siegfried, 1997).
In this study, the role of class size in affecting different types of student achievement was examined including separately analyzed student performance on the three components of the TUCE III score: (a) "RU," recognition and understanding of basic terms and principles, (b) "EA," explicit application of basic terms, concepts and principles, and (c) "IA," implicit application of basic terms, concepts, and principles. Based on the literature (Raimondo, Esposito, & Gershenberg, 1990), it was expected that any influence class size may have should manifest itself in EA or IA. The student performance data represented 2,143 students, 1,527 of whom had SAT scores available (Kennedy & Siegfried, 1997).
Two major findings resulted from this study. The first was that once proper control is made for ability, none of the variables in this particular data set, over which the instructors or department chairs had control, influence achievement. Secondly, the final conclusion of this study is that large class size does not reduce the ability to learn the principles of economics (Kennedy & Siegfried, 1997).
These findings are among the most convincing in this review due to the fact that such careful consideration was applied to the measurement of variables. Also, several types of institutions were represented, and class size was fairly evenly distributed over a wide range, adding strength to the study's methodology. However, Kennedy and Siegfried (1997) do state that one reason for this result is that instructors of economics classes do not change their behavior as class size changes. And furthermore, as Hansen (1970) has suggested, the "no significant difference" conclusion is reached frequently in education studies because any given technique may affect some students positively and others negatively, with no difference in the class as a whole. Kennedy and Siegfried (1997) warn that results from studies such as this should not be interpreted as a call to throw all students into one gigantic class as a cost saving measure, but to offer a mixture of different types of classes and allow students to select the class size that best suits their learning style.
A study conducted by Hou (1994), in which class size and determinants of learning effectiveness in economics were examined, utilized a highly quantitative approach to every aspect of the research. As a result, the study is mired in statistical formulas and difficult for one without expertise in the field of statistics to comprehend with ease. This study used a technique originally designed for discrimination analysis to decompose the measured learning differences in managerial economics (in terms of test process) into a portion that is accountable by the differences in student characteristics, and the residual portion, which can be interpreted as the difference due to class size. Achievement was defined in terms of a regression equation. Class size was defined in terms of "regular" (n = 25) and "oversized" (n = 54) classes. (Hou, 1994).
Surprisingly, results of this study indicate that significant superior performance was found in favor of the large class. Additionally, it was found that overall GPA and performances in previous college economics courses do not have any significant impact on the current score, although Hou (1994) does caution the readers that due to the size of the sample, the robustness of some of the findings are suspect.
A qualitative study conducted by Hofmann, Posteraro, and Presz (1994) examined (a) the college graduate's perceptions relative to what facilitated his/her continued enrollment and ultimate success in graduating, (b) the college graduate's perceptions of adult learner needs and ways the college supported those needs, and (c) how the college could improve its supportive efforts. The sample consisted of 39 recent graduates of a small liberal arts college in New England across three educational programs. The participants were interviewed via telephone with a semi-stratified phone interview. Simultaneously, a faculty forum was also conducted to address the same issues (Hofmann et al., 1994).
One of the most interesting results was that graduates cited faculty as the most important factor contributing to their success in obtaining their degree, noting that faculty was the major way in which the college contributed to their success. Another desirable feature noted was the small class size (Hofmann, Posteraro, & Presz, 1994). This finding is important to note in that it illustrates students' perceived benefits of instructor interaction and small class size.
A final article by Kopeika (1992) is anecdotal in nature, and reports an electrical engineering department's view on the relationship of smaller class sized to academic retention, success, and perceived quality of training as reported by subsequent employers. Kopeika reports that beginning in 1985-1986, all compulsory departmental courses were split into two more-or-less equal classes of 65-70 students each in order to reduce class size. Sample characteristics were merely described as students pursuing a B.Sc. in the department of electrical and computer engineering at the university.
Kopeika (1992) asserts that when class size was reduced, there was an increase of 30 students in the number of graduating students that year. Additionally, feedback from industry and graduate schools indicated that the quality of the students' academic level had increased, even though the preparedness of incoming students had decreased. Kopeika asserts that the increased number of graduating students provides a revenue increase that more than offsets the increase in teaching expenses associated with smaller class sizes. (Kopeika, 1992).
SUMMARY AND RECOMMENDATIONS FOR FUTURE RESEARCH
It has been stated previously in this review that research on class size and achievement in higher education remains inconclusive. In this paper, eight very methodologically diverse studies were discussed. The results were as varied as the methods, with two studies showing no relationship between class size and achievement, three indicating a negative relationship, two showing mixed results, and another reporting a positive relationship between these two variables. Additionally, a common theme appears to emerge from many of these studies which is a lack of statistical power due to small samples, loose overall methodological design, narrow assessments of achievement as measured by course grade alone, and in some cases, a lack of quantitative data.
With reference to these limited findings, it is clear that more statistically and methodologically powerful studies need to be conducted if there is to be any insight into this issue. Another suggestion is to consider additional factors that may be confounding the issue of class size and achievement. For example, Noble (2000), in examining the assumptions about the possible cause-effect relationship between class size and student achievement, also focused on areas related to the improvement of student achievement. These areas include (a) student aptitude and readiness for undergraduate and graduate school education, (b) remediation for ill-prepared and educationally disadvantaged students, (c) an understanding of learning styles and processes as well as teaching styles and mind-leading skills, (d) student motivation and
effort, and (e) widespread grade inflation and the "watering down" of standards. (Noble, 2000.)
In an attempt to gain further insight, it appears to be of utmost importance that the issues identified by Noble (2000) be further explored in future studies. Secondly, it is recommended that researchers attempt to develop more comprehensive methods for assessing achievement, as opposed to the prevalent limited measure of course grade alone. Third, the possibility of a ceiling effect in grading practices and the lowering of academic standards should inspire vigilance in educators, administrators, and researchers in institutions of higher education. Finally, as more research is published that addresses these concerns, a meta-analysis should be conducted to provide a more comprehensive and integrated approach to the issues of class size and achievement in higher education. Such an analysis could provide the information needed to make decisions which will ensure a high quality of academic standards in our institutions of higher education, thus promoting the development of the well-prepared and knowledgeable leaders needed in our complex society at the new millennium.
Borden, V. M. H., & Burton, K. L. (1999, May). The impact of class size on student performance in introductory courses. Paper presented at the Annual Forum of the Association for Institutional Research, Seattle, WA. (ERIC Document Information Service No. ED 433 782)
Gibbs, G., Lucas, L., & Simonite, V. (1996). Class size and student performance: 1984-94. Studies in Higher Education, 21, (3), 261-273.
Hancock, T. M. (1996). Effects of class size on college student achievement. College Student Journal, 30, 479-481.
Hansen, L., Kelly, A., and Weisbrod, B. (1970) Economic efficiency and the distribution of benefits from college instruction. American Economic Review: Papers and Proceedings, 60, 364-369.
Hofmann, J. M., Posteraro, C., & Presz, H. A. (1994, May). Adult learners: Why were they successful? Lessons learned via an Adult Learner Task Force. Paper presented at the Adult Learner Task Force Conference, Columbia, SC. (ERIC Document Service No. ED 375 269)
Hou, J. W. (1994). Class size and determinants of learning effectiveness. (ERIC Document Reproduction Service No. ED 377 239)
Kennedy, P. E., & Siegfried, J. J. (1997). Class size and achievement in introductory economics: Evidence from the TUCE III data. Economics of Education Review 16, (4), 385-394.
Kopeika, N. S. (1992). On the relationship of number of students to academic level. IEEE Transactions on Education. 35, (1), 89-102.
Noble, Jr., J. H. (2000). Cherchez l'argent: A contribution to the debate about class size, student faculty ratios, and the use of adjunct faculty. Journal of Social Work Education. 36, (1), 89-102.
Raimondo, H. J., Esposito, L., & Gershenberg, I. (1990). Introductory class size and student performance in intermediate theory courses. Journal of Economic Education, 21, (4), 369-381.
Van Allen, G. H. (1990). Educational morality: A task of resisting the economic corruption of academic excellence. (ERIC Document Reproduction Service No. ED 317 232)
The authors wish to thank Dr. Richard Helldobler, Professor Sam Lonich and Dr. Dennis Sweeney for their support of this project.
LINDA S. TOTH AND LINDA G. MONTAGNA Psychology Department California University of Pennsylvania California, PA 15419…
Questia, a part of Gale, Cengage Learning. www.questia.com
Publication information: Article title: Class Size and Achievement in Higher Education: A Summary of Current Research. Contributors: Toth, Linda S. - Author, Montagna, Linda G. - Author. Journal title: College Student Journal. Volume: 36. Issue: 2 Publication date: June 2002. Page number: 253+. © 2009 Project Innovation (Alabama). COPYRIGHT 2002 Gale Group.