Accounting Education ISSN: 0963-9284 (Print) 1468-4489 (Online) Journal homepage: https://www.tandfonline.com/loi/raed20 Active learning in small and large classes Brock Murdoch & Paul W. Guy To cite this article: Brock Murdoch & Paul W. Guy (2002) Active learning in small and large classes, Accounting Education, 11:3, 271-282, DOI: 10.1080/0963928021000031448 To link to this article: https://doi.org/10.1080/0963928021000031448 Published online: 05 Oct 2010. Submit your article to this journal Article views: 317 View related articles Citing articles: 1 View citing articles Full Terms & Conditions of access and use can be found at https://www.tandfonline.com/action/journalInformation?journalCode=raed20 Accounting Education 11 (3), 271–282 (2002) Active learning in small and large classes B RO C K M UR D O CH * an d PAU L W. GU Y California State University, Chico, USA Received: July 2001 Revised: January 2002; June 2002 Accepted: July 2002 Abstract This research investigates the effect of class size on introductory accounting student performance within the context of an active learning environment. In-class group activities were implemented as an integral part of the learning environment in both small and large sections. Although the class size issue has been investigated before, this paper focuses on whether active learning methods are differentially effective in large and small classes when learning is measured by performance on exams emphasizing analytical problems and essay questions. Because practical and ethical reasons prevented students from being randomly assigned to large and small sections, the research methodology controls for confounding in uences. Speci cally, we control for the covariates age, attendance, gender, grade point average, and homework completion. Small class students scored signi cantly higher on the nal exam than did students in the large section. Keywords: active learning, class size, covariates, critical thinking skills, group activities Introduction This paper reports on a research project which investigates the effect of class size on Introductory Accounting students’ performance within the context of an active learning environment at a university in the USA. Although the class size issue has been investigated before, this paper focuses on whether active learning methods are differentially effective in large and small classes when learning is measured by performance in examinations emphasizing analytical problems and essay questions. Accounting educators should not be convinced that results showing equal performance on multiple choice examinations for small and large sections, when only the lecture method is used, can be generalized to learning environments in which group activities are used extensively and examinations require analytical reasoning, decision-making, and explanatory written responses. Literature review The issue of class size and its impact on student learning has been a long-standing controversy across disciplines and education levels (Siegel et al., 1959; Simmons, 1959; Shane, 1961; Laughlin, 1976; McConnell and Sosin, 1984; Williams et al., 1985). Conclusions have varied across disciplines. Simmons (1959) observed a higher failure rate and lower overall achievement for intermediate algebra students in large class settings. * Address for correspondence: Dr. Brock Murdoch, Professor of Accounting, California State University, Chico, College of Business, Department of Accounting and MIS, Chico, CA 95929-0011, USA. E-mail: bmurdoch@csuchico.edu Accounting Education ISSN 0963–9284 print/ISSN 1468–4489 online © 2002 Taylor & Francis Ltd http://www.tandf.co.uk/journals DOI: 10.1080/096392802100003144 8 272 Murdoch and Guy However, Siegel (1959), Laughlin (1976), and Williams et al. (1985), in analyses across many disciplines, all concluded that class size does not affect student learning. On the issue of accounting class size, Anderson (1964) argued that, while small classes with excellent teachers are optimal, large classes with excellent faculty are better than small classes with less quali ed teachers. A study by Baldwin (1993) focused on this very issue. It investigated whether an established teacher with an excellent reputation as a lecturer could achieve results in a mass lecture section that were equal to or better than doctoral students could achieve with smaller sections. Baldwin concluded that there was no signi cant difference in students’ performance. Similarly, Hill (1998) found no performance advantages for small classes. Indeed, when Hill controlled for grade point average (GPA) and attendance, the large section outperformed the small section in both nal examination score and overall course grade. Hypothesis development The Baldwin (1993) study has limitations that suggest additional research may add to the knowledge regarding class size in accounting courses. In his research, Baldwin admitted that an award-winning teacher taught the large section while doctoral students taught the smaller sections and that this instructor ‘perceived some pressure to do a good job so as not to injure an otherwise ne teaching reputation’ (p. 110). Baldwin’s objective was not to control for variables other than size, but to compare results from Introductory Accounting taught in a mass-lecture format by an outstanding teacher to results obtained by doctoral student instructors in smaller sections. Finally, Baldwin states, ‘Another limitation common to this type of experimentation is that the outcome assessment tool used was a set of multiple choice examinations’ (1993, p. 110). Hill (1998) also acknowledged limitations of basing measures of performance on multiple choice examinations and the lecture method. She recognized that ‘multiple choice examinations do not allow instructors to evaluate high-order cognitive skills . . .’ (p. 62). Additionally, Hill indicated that alternative pedagogies, such as cooperative learning and group activities, were not considered feasible by the instructors participating in her research. Consequently, the lecture method was used in all sections. This research examines issues similar to those investigated by Baldwin and Hill in a different learning and testing environment. In contrast to the Baldwin and Hill studies, how class size interacts with an active learning environment (i.e., group activities) to affect performance on examinations emphasizing critical thinking skills is examined. It might be argued that group cases can be used more effectively in small classes to enhance critical thinking skills because the instructor has more time to devote to both groups and individual students within groups. Consequently, superior performance is expected in smaller classes on examinations consisting of analytical problems and essays. There is evidence that multiple-choice questions fail to distinguish between different levels of ability (Rogers and Bateson, 1991; Phillips, 1999; Rayburn and Rayburn, 1999; Burton, 2001). Both Baldwin and Hill acknowledged that the ability to generalize their ndings is limited by their exclusive use of multiple-choice examinations. It is maintained that the ability of Baldwin and Hill to generalize their ndings, as acknowledged by Hill, is also limited by reliance on the lecture method. Empirical data are lacking with respect to the relationship between class size and performance, within the context of group activities, when examinations are developed to evaluate critical thinking Active learning in small and large classes 273 skills. Both the general and accounting literature on class size, however, takes the position that small classes enhance learning and critical thinking (Simmons, 1959; Schattke and McAllister, 1962; Anderson, 1964; McCormick, 1967; McKeachie, 1970; Kempner, 1970; McConnell and Sosin, 1984). Arguments that active learning methods result in improved learning and critical thinking skills are persuasive (Christensen, 1991; Garvin, 1991; Wilkinson and Dubrow, 1991; Cottell and Millis, 1993). Intuitively, it would be expected that the instructor in smaller classes is more able to facilitate group interaction, resulting in more stimulating discussions of analytical and theoretical issues. To the extent that such issues are tested in greater depth by problem and essay examinations, students who have learned more have the opportunity to exhibit this deeper knowledge. While active learning techniques have been linked to superior learning (Rau and Heyl, 1990; Burns and Mills, 1997), it is not clear that group activities, speci cally, are effective in large class settings. It would be expected that group cases is less common in large classes due to the increased need for the instructor to spend time with each group. Consistent with the foregoing arguments, this study predicts that students in small classes will perform better than students in large classes in examinations that include analytical problems and essay questions. It is believed that these types of examinations more effectively test critical thinking skills. Consequently, it is expected to nd evidence to support the alternative hypothesis below: Ha: Students in small classes emphasizing group activities will perform better on analytical problems and essay questions than students in large classes emphasizing group activities. The null hypothesis is that there is no performance difference between the small and large groups. Description of the study In the early 1990s, one campus in a state university system in the USA reengineered its rst year of Accounting with a grant from the US Department of Education’s Fund for the Improvement of Postsecondary Education (FIPSE). These rst two semester courses, as reengineered, are predominantly user-oriented, a change from the prior traditional preparer-oriented rst year of Accounting. The reasons for the change were related to the Accounting Education Change Commission’s (1990) call for universities to move away from the rule-based and procedure-oriented focus of most Introductory Accounting courses to curricula that emphasize thinking critically, solving unstructured problems, making decisions, and communicating effectively. Five sections of the rst semester of Introductory Accounting were offered in the Fall of 1998. This course is a mix of nancial and managerial accounting concepts and was designed to focus on the aforementioned critical thinking, problem-solving, decisionmaking, and communicating skills. Of the ve sections offered, four were smaller classes, with enrolments of 38, 39, 40, and 37. The other was a large section with 280, for a total of 434 originally enroled. Although it would have been preferable to have assigned students randomly to small and large classes, practical and ethical considerations prevented such a process. Students chose which section of the course in which to enrol. There were 274 Murdoch and Guy a few transfers, additions, and deletions involving all sections that occurred during the rst two weeks of the course.1 The large section met on Mondays, Wednesdays, and Fridays (MWF) from 10:00 to 10:50 am. The four small sections met (i) MWF from 11:00 to 11:50 am, (ii) Tuesdays and Thursdays (TTh) from 11:00 am to 12:15 pm, (iii) TTh from 12:30 to 1:45 pm, and (iv) TTh from 2:00 to 3:15 pm. Identical course content, syllabus, assignments, and scheduled activities were used in both small and large sections. Cases were extensively used to promote discovery learning. The cases were decision-making ones in which students were given both relevant and irrelevant data. Students worked in groups during class to analyse these data in order to make a decision. For example, an early case asked students to decide whether a bank should make a short-term business loan when the borrower had considerable equity, but little in the way of liquid assets. A later case required students to determine, for a new rm, which was the better inventory method, FIFO or LIFO, for purposes of supporting a bank loan request and for maximizing cash ow. Instructors in the small sections circulated during the time allocated to these group activities, providing clari cation and assuring that groups stayed on task. The large section had several student mentors (upper division Accounting majors) who assisted the large section instructor, since the one instructor could not monitor all groups. There were 15 group cases during the 15-week semester. Fourteen of these cases were done during class time, with each student individually responsible for reading the case, and sometimes completing preliminary work, prior to coming to class. The sole group case to be done outside of class time was the preparation of a small business plan, including pro forma nancial statements. Each group was responsible for presenting its business plan to the rest of the class. Most students were randomly assigned by the instructor to a group of ve. A few groups were assigned a minimum of four or a maximum of six students because the number of students in each section was often not divisible by ve. Approximately 40% of all class time was set aside for students to work in group activities and to get feedback on these activities. The course had an overall emphasis on student and instructor interaction, with each instructor requiring students to defend their individual as well as group decisions. Graders were provided to all instructors to grade both individual and group assignments. Total points allocated to group activities and individual homework assignments were 230 and 270, respectively. Group members were required to attend on the day the group case was done in class to receive points for that case. The mid-term and nal examinations were each worth 200 points. Course grades were assigned by the individual instructor based on the points earned by each student out of a possible 900. At the end of each of four learning modules, each student lled out an evaluation form for everyone in the group, including herself or himself. These evaluations were used by the instructor to reward or penalize individual group members, if necessary, so that more deserving group members received more points. Point reallocations were very unusual, however. Judging by the group evaluation forms, most students worked reasonably hard in 1 Similar to most US universities, ‘shopping for classes’ occurs during the rst few class periods. So as to include only serious students in the study, of cial rolls at the end of the second week of classes were used as the sample. Active learning in small and large classes 275 their groups. Most differential evaluations re ected differences in student ability rather than differences in effort. We believe the real value of the students’ peer evaluations was that they motivated students to participate. Some additional controls were added as well. Speci cally, this study includes the following characteristics and controls that differentiate it from Baldwin (1993) and Hill (1998): c c c Experienced tenured faculty, familiar with the course, taught all sections. Examination questions were known only to the researchers prior to administering the examinations to students. Instructors were not involved in examination preparation or grading. Grading of examinations was the responsibility of a professor emeritus with considerable experience in grading academic and professional examinations (i.e., the CPA and CMA examination). The grader did not know which examinations were from the small sections and which were from the large section. Method Data collection Three different male faculty members of similar experience were used to teach the ve different sections of the course. Indeed, all three faculty members who taught the course were part of the original FIPSE grant team who developed course materials in the early 1990s. Student evaluations from prior semesters for these faculty for the same course were examined and did not indicate any signi cant differences among them. Attendance and homework data were recorded for all students. Student assistants collected and evaluated homework assignments as to completeness. Assignments were not graded as to correctness, since this type of evaluation may have measured the same type of aptitude represented by GPA (discussed later as a covariate). Students were given full credit if their homework was complete and submitted on time. Performance measures were obtained from a nal examination administered to all students at the same time. At the research university, mid-term examinations are administered in regular class periods at the instructor’s discretion while nal examinations occur at the same time for all students enroled in the course. The researchers tried to implement special procedures to provide for a common mid-term examination so that information leakage from earlier mid-terms to later ones would not be an issue. However, a large proportion of students were unable to take the mid-term examinations at the common examination time. Consequently, performance measures are based solely on the nal examination. Dependent variable The dependent variable is the correct score, out of 164 non-multiple choice points, for the nal examination. Problems in the nal examination required each student to choose relevant information from an array of data, perform analysis, and interpret or explain the meaning of his or her analysis. 276 Murdoch and Guy Independent variables The independent variable of interest in this investigation is class size. However, since it was not possible to select students randomly into large and small classes, it is necessary to control for other variables that may confound the effect class size has on nal examination score. Potential confounding in uences for which we controlled are aptitude (GPA), motivation (attendance and homework completion), gender, and age. GPA at the research university 2 is used as the measure of aptitude. Students who have done well in other academic courses are expected to do well in Accounting (Wooten, 1998). Both Baldwin (1993) and Hill (1998) found GPA to be highly correlated with examination performance. Several researchers have studied the effect of gender on performance, often with contrary results. Mutchler et al. (1987) concluded that females outperformed males and that female-instructed Accounting students outperformed male-instructed students. Lipe (1989), however, was unable to replicate these ndings, while Doran et al. (1991) found that males scored signi cantly higher in examinations than females. Tyson (1989) concludes that any superiority of female performance in Accounting is related to females’ higher work needs and not to ‘extrinsic or accounting-related factors’ (p. 159). Similarly, Buckless et al. (1991) concluded using scholastic aptitude test (SAT) or American College Testing Assessment Test (ACT) as a covariate reduces the gender effect related to examination scores to insigni cance. Hill (1998) determined class attendance to be signi cantly correlated with examination performance in the rst Introductory Accounting course. Both attendance and homework completion are considered by the authors to be good measures of student motivation. Baldwin (1993) and Hill (1998) both included student’s age as a potential covariate, but only Baldwin found age to be ‘signi cantly associated with performance’ (p. 102), ostensibly because it measures maturity, which may be a substitute for motivation. Statistical analysis As discussed above, confounding variables that may affect performance were controlled for, exclusive of size. The following regression is used to predict students’ scores on the nal examination: Final exam score = b 0 + b 1Size + b 2Age + b 3Att + b 4Gender + b 5GPA + b 6HW where Final exam score = Points correct on nal examination Size = 0 for small section and 1 for large section Age = Age in years Att = Percentage of class periods attended Gender = 0 for female and 1 for male GPA = Grade point average (4 = A, 3 = B, 2 = C, 1 = D, 0 = F) HW = Percentage of total homework assignments completed A one-tail t-test for the size variable is used to determine the signi cance of the size variable in the regression. 2 GPA at the university where the research was conducted, rather than overall GPA, was used because of potential institutional grading differences among the many colleges and universities (mostly community colleges) from which students transfer. 277 Active learning in small and large classes Results Table 1 displays the numbers of students who were eliminated from statistical tests and the reason for elimination. Of the 434 original enrolments, a total of 355 students took the nal examination, representing an overall completion rate of 81.8% (355 of 434 students). The small sections had a completion rate of 79.9% (123 of 154 students) and the large section had a completion rate of 82.8% (232 of 280 students). Although course completion, (or its complement, attrition) is not the focus of this study, it is important to discern whether course completion differs between large and small sections. Signi cantly different course completion might indicate undetected differences in the two samples. Table 1. Sample size Number of students: All % Small % Large % Originally enroled Who did not complete the course Taking nal examination (completion rate) In rst semester with no GPA Used in hypothesis test 434 79 355 100.0 18.2 81.8 154 31 123 100.0 20.1 79.9 280 48 232 100.0 17.2 82.8 99 256 22.8 59.0 41 82 26.6 53.3 58 174 20.7 62.1 A student was considered to have not completed the course if he or she was enroled after the second week of the course, but did not take the nal examination, regardless of the reason. University rules require those who of cially drop out after the rst two weeks to have a serious and compelling reason to do so. Examples of serious and compelling reasons are illness, family crisis, nancial hardship, and similar excuses not related to course performance. Poor performance is not considered a serious and compelling reason. A chi-square test showed there is no evidence that course completion differed between small and large sections.3 Sample size was reduced further because rst semester students did not have a GPA from the research university, a necessary independent variable. This eliminated 99 of 355 students. Data from the surviving 256 students in the sample were included in hypothesis tests. Preliminary statistics Table 2 displays preliminary statistics (means and proportions) for the dependent variable ( nal examination score) and independent variables (age, attendance, gender, GPA, and homework completion percentage) for the small sections and the large section. A one or two-tail p-value for the signi cance of the difference between the small and large means (t-tests) or proportions (chi-square) is also displayed, as is the conclusion as to whether the difference is statistically signi cant. A one-tail test is used when the a priori expectation is that the appropriate characteristic would be greater for the small sections. A two-tail test 3 A chi-square of 0.596 was calculated with a corresponding p-value of 0.44. 278 Murdoch and Guy Table 2. Preliminary statistics: means/proportions for small and large sections Variable Small = S (n = 82) Large = L (n = 174) A priori Expectation* p = value** Conclusion Final exam score 91.39 78.63 m S> m L 0.0013 Age 22.65 20.76 m S¹ m L 0.0003 Attendance % 88.81 84.58 m S> m L 0.0075 Gender (male %) 54.88 GPA 2.756 58.62 2.597 pS ¹ pL m S¹ m L 0.5720 0.0497 HW completion % 76.73 m 0.1120 Final exam score Age Attendance % Gender (male %) GPA HW completion % 79.69 S> m L Small classes scored higher Small class students were older Small classes had better attendance No signi cant difference Small class students had higher GPA No signi cant difference Points correct on nal examination Age in years Percentage of class periods attended Percentage of male students in the sample Grade point average (4 = A, 3 = B, 2 = C, 1 = D, 0 = F) Percentage of total homework assignments completed *The main a priori hypothesis is that students in small classes have higher nal exam scores. For the covariates attendance and homework, the a priori expectation is that students in small classes will attend more frequently and complete more homework. For the covariates age, gender, and GPA, the a priori expectation is that there may be a difference between small and large classes (due to a lack of random assignment of students to sections) without any conjecture as to the direction of the difference. ** Final exam score, Age, Attendance %, GPA, and HW completion % represent means and the p-values are from ttests. Gender (Male %) is a proportion and the p-value is from a chi-square test. is used when the expectation is that there may be a difference between small and large sections, but the direction of the difference is not clear a priori. Table 3 is a matrix showing the correlations among all variables. In all cases, the signs of correlations between the dependent and independent variables are consistent with expectations, given prior research and/or theoretical considerations. If independent variables are highly correlated, a condition called multicollinearity, the regression coef cients can become unstable and the signi cance tests and the interpretations of the individual variables may be misleading (Neter et al., 1996, pp. 289–90). The correlations among the independent variables are small or moderate, with the largest correlations being .523 between HW completion and attendance, and .501 between HW completion and GPA. To see if the independent variables in combination might show multicollinearity, for each independent variable we computed a diagnostic statistic called the variance in ation factor (VIF) (Neter et al., 1996, pp. 386–87). If the independent variables are uncorrelated (a rare situation in practice) the VIFs will be 1. It is suggested that VIFs above 10 could indicate serious multicollinearity. However, the largest VIF here is 1.613 (for HW) and the main variable of interest, size, has a VIF of 1.091. Consequently, multicollinearity is not a problem. 279 Active learning in small and large classes Table 3. Correlation matrix (n = 256) Final exam score Size Age Attendance Gender GPA Final exam score Size Age Attendance % Gender GPA HW Completion % 1 –0.187 0.136 0.247 –0.001 0.399 0.398 1 –0.223 –0.152 0.035 –0.123 –0.076 1 –0.048 0.111 –0.038 –0.056 1 –0.161 0.387 0.523 Final exam score Size Age Attendance % Gender GPA HW completion % Points correct on nal examination 0 for small section and 1 for large section Age in years Percentage of class periods attended 0 for female and 1 for male Grade point average (4 = A, 3 = B, 2 = C, 1 = D, 0 = F) Percentage of total homework assignments completed 1 –0.216 –0.127 1 0.501 HW Completion % 1 Hypothesis test and conclusion The hypothesis concerns student performance in examinations consisting of analytical problems and essay questions. The one-tail alternative hypothesis is reproduced below: Ha: Students in small classes emphasizing group activities will perform better on analytical problems and essay questions than students in large classes emphasizing group activities. Partial regression coef cients, a priori expected coef cient signs, t-statistics, and pvalues for each of the six independent variables from the regression are reported in Table 4. These six variables combined explain 25.4% of the variation in nal examination score. The t-value for size is –1.876 with a one-tail p-value of .03. Accordingly, it is concluded that utilizing group learning techniques in small sections results in a higher level of learning than utilizing these same techniques in large sections. The independent variable of interest, size, has a partial regression coef cient of –7.301. This indicates that for same sex students with equal age, attendance, GPA, and HW, the nal examination scores in the small sections would be predicted to be an average of more than 7 points higher. Although, small sections students scored an average of over 12 points higher on the nal examination (see Table 2), only about 7 points of this difference is associated with smaller class size. Because multicollinearity is not an issue (see discussion above), there are no concerns about the conclusion of the hypothesis, nor with the foregoing statement about how class size affects nal examination score. Limitations Common limitations related to conducting student performance research are also present here, although the authors attempted to control for these. This research was conducted at 280 Murdoch and Guy Table 4. Regression for predicting nal examination points Regression: Final exam score = b n = 256 R2 = 0.254 0 + b 1Size + b 2Age + b 3Att + b 4Gender + b 5GPA + b 6HW Independent variable Coef cient Intercept Size Age Att (attendance %) Gender (male %) GPA HW (completion %) –12.702 –7.301 1.030 0.011 5.308 14.248 0.472 A priori Expectation* t-statistic p-value b –1.876 2.255 0.069 1.459 4.163 3.871 0.031 0.012 0.473 0.146 0.000 0.000 1< 0 > 2 0 3>0 4¹ 0 5>0 6>0 b b b b b ANOVA Regression Residual Total Final exam score Size Age Attendance % Gender GPA HW completion % Df SS MS F p-value 6 249 255 65 639.5 192 850.7 258 490.2 10 939.9 774.5 14.125 0.000 Points correct on nal examination 0 for small section and 1 for large section Age in years Percentage of class periods attended 0 for female and 1 for male Grade point average (4 = A, 3 = B, 2 = C, 1 = D, 0 = F) Percentage of total homework assignments completed * The main a priori hypothesis is that students in small classes have higher nal exam scores. For the covariates age, attendance, GPA, and homework, the a priori expectation is that older students with better attendance, higher GPAs, and who complete more homework score higher in the nal exam. There is no a priori expectation as to which gender scores higher in the nal exam. one university, in one semester, and for one course in one country. Since students could not be randomly assigned to classes, aptitude and motivation differences are real possibilities. Accordingly, the research method controlled for these differences, as well as for gender and homework completion (another measure of motivation). Conclusions and suggestions for future research It was hypothesized in this research that group learning techniques could be used more effectively in small classes than in large and would result in superior examination performance in small classes. This hypothesis is supported by the research. Although this nding may suggest that group activities should only be used in small classes, it does not directly address the issue of whether group activities may also improve Active learning in small and large classes 281 learning in large classes. Past research has shown that class size does not affect performance when using the lecture method and objective examinations (Baldwin, 1993; Hill, 1998). Comparing the use of active-learning methods (e.g., group activities) in large classes to the lecture method in large classes, coupled with problem and essay examinations, will provide further evidence of the effectiveness of active learning methods. Is the often cited belief that large classes preclude the effective use of active learning methods warranted? Addressing other research issues regarding the effect of group activities on performance is also recommended. Some possibilities are examining (i) the effectiveness of group activities in small and large classes with only multiple choice testing, (ii) the effectiveness of group activities without awarding points for participation, and (iii) the effectiveness of student peer evaluations. Acknowledgements The authors would like to thank Tony Catanach, Paul Krause, Richard Lea, and two anonymous reviewers for their useful comments on this paper. References Accounting Education Change Commission (1990) Objectives of education for accountants: Position Statement Number 1. Issues in Accounting Education 5(1), 307–12. Anderson, W.T. (1964) Large classes in elementary accounting. The Accounting Review 39(4), 1034–36. Baldwin, B.A. (1993) Teaching introductory nancial accounting in mass-lecture sections: Longitudinal evidence. Issues in Accounting Education 8(1), 97–111. Buckless, F.A., Lipe, M.G. and Ravenscroft, S.P. (1991) Do gender effects on accounting course performance persist after controlling for general academic aptitude? Issues in Accounting Education 6(2), 248–61. Burns, C.S. and Mills, L.K. (1997) Bringing the factory to the classroom. Journal of Accountancy 183(1), 56, 58, 60. Burton, R.F. (2001) Quantifying the effects of chance in multiple choice and true/false tests: question selection and guessing of answers. Assessment & Evaluation in Higher Education 26(1), 41, 50. Christensen, C.R. (1991) Premises and practices of discussion teaching. Education for Judgment, pp. 15–34. Boston, MA: Harvard Business School Press. Cottell, P.G. and Millis, B.J. (1993) Cooperative learning structures in the instruction of accounting. Issues in Accounting Education 8(1), 40–59. Doran, B.M., Bouillon, M.L. and Smith, C.G. (1991) Determinants of student performance in Accounting Principles I and II. Issues in Accounting Education 6(1), 74–84. Garvin, D.A. (1991) A delicate balance: ethical dilemmas and the discussion process. Education for Judgment, pp. 287–303. Boston, MA: Harvard Business School Press. Hill, M.C. (1998) Class size and student performance in Introductory Accounting courses: further evidence. Issues in Accounting Education 13(1), 47–64. Kempner, J.J. (1970) An argument for small class size. The Accounting Review 45(2), 364–66. Laughlin, J.S. (1976) A sacred cow – class size. College and University 51(3), 339–47. Lipe, M.G. (1989) Further evidence on the performance of female versus male accounting students. Issues in Accounting Education 4(1), 144–52. 282 Murdoch and Guy McCormick, F.L. (1967) Large group instruction in elementary accounting. The Accounting Review 42(3), 592. McConnell, C.R. and Sosin, K. (1984) Some determinants of student attitudes toward large classes. Journal of Economic Education 15(3), 181–90. McKeachie, W.J. (1970) Research on College Teaching: A Review. Washington, DC: ERIC Clearinghouse on Higher Education cited in E.K. St. Pierre, M.P. Riordan and D.A. Riordan (eds) 1990, Research in Instructional Effectiveness, p. 138. Harrisonburg, VA: Center for Research in Accounting Education, School of Accounting, James Madison University. Mutchler, J.F., Turner, J.H. and Williams, D.D. (1987) The performance of female versus male accounting students. Issues in Accounting Education 2(1), 103–11. Neter, J., Kutner, M.H., Nachtsheim, C.J. and Wasserman, W. (1996) Applied Linear Regression Models. Third Edn. Chicago: Richard D. Irwin, Inc. Phillips, F. (1999) Business students’ learning preferences and associated task performance. Journal of Education for Business 75(1), 27–32. Rau, W. and Heyl, B.S. (1990) Humanizing the college classroom: collaborative learning and social organization among students. Teaching Sociology 18, 141–55. Rayburn, L.G. and Rayburn, J.M. (1999) Impact of course length and homework assignments on student performance. Journal of Education for Business 74(6), 325–31. Rogers, W.T. and Bateson, D.J. (1991) The in uence of testwiseness upon performance of high school students on school leaving examinations. Applied Measurement in Education 4(2), 159–83. Schattke, R. and McAllister, L. (1962) Large versus small classes in elementary accounting. The Accounting Review 37(3), 557–61. Shane, H.G. (1961) What research says about class size and human development. NEA Journal 50(1), 30–32. Siegel, L., Macomber, F.G. and Adams, J.F. (1959) The effectiveness of large group instruction at the university level. Harvard Educational Review 29(3), 216–26. Simmons, H.F. (1959) Achievement in intermediate algebra associated with class size at the University of Wichita. College and University 34(3), 309–15. Tyson, T. (1989) Grade performance in Introductory Accounting courses: why females outperform males. Issues in Accounting Education 4(1), 153–60. Wilkinson, J. and Dubrow, H. (1991) Encouraging independent thinking, Education for Judgment. Boston, MA: Harvard Business School Press, pp. 249–61. Williams, D.D., Cook, P.F., Quinn, B. and Jensen, R.P. (1985) University class size: is smaller better? Research in Higher Education 23(3), 307–18. Wooten, T.C. (1998) Factors in uencing student learning in introductory accounting classes: a comparison of traditional and nontraditional students. Issues in Accounting Education 13(2), 357–73.