Group versus individual learning of quantitative accounting topics: effects on test performance in the first-year accounting course.

Cagwin, Douglass ; Barker, Katherine J.

ABSTRACT

Educators continue to search for ways to improve both accounting and methods of teaching. Increased use of cooperative learning is often a feature of curriculum revision. Although previous research has shown that cooperative learning techniques can sometimes lead to improved student learning, there has been no research that has examined the effects of specific cooperative techniques (e.g., group homework assignments) on learning specific quantitative business topics.

This is a field study of first-year accounting students at a large Southeastern university. Multiple regression analysis is used to determine whether there is a difference in test performance on quantitative accounting topics between students completing graded homework in groups versus students completing the same assignments individually.

The results of this field experiment indicate that test performance of two specifically targeted quantitative topics was not influenced by using the cooperative learning technique of graded group assignments. Therefore business instructors may feel free to use this cooperative learning technique without fear that it may jeopardize learning quantitative topics. This research did find a positive relationship between quantitative test performance and a higher number of university credit hours completed prior to exposure of the tested quantitative topics. This finding may help to guide those charged with revising business curriculum to introduce quantitative accounting topics later rather than earlier in the sequencing of required business courses.

INTRODUCTION

For nearly two decades there have been many appeals from both accounting professionals (e.g., American Accounting Association [AAA], 1986); Arthur Andersen et al., 1989; and Accounting Education Change Commission [AECC], 1990, 1992) and academics to improve undergraduate accounting education; yet the debate continues as to how the accounting curriculum or methods of teaching should be revised. Efforts by business schools and individual business disciplines to improve the manner in which courses are delivered have included an assortment of educational methods (e.g., case studies, group projects, in-class projects, cooperative learning assignments, and community service learning projects). The use of cooperative learning techniques has often been a feature of curriculum revision, particularly since many employers have embraced a more cooperative focus in the workplace.

Cooperative learning has been defined by Cooper, et al. (1990) as: "An instructional technique which requires students to work together in small fixed groups on a structured learning task." Previous research has shown that the use of cooperative learning techniques generally, but not always, leads to increased learning by students. An underlying assumption is that by working together, students will help teach each other (Gilbert-MacMillan, 1983; Parker, 1984). However, little is known about the effects of cooperative techniques in specific learning situations (e.g., group vs. individual homework assignments) or with regard to learning specific quantitative accounting material. Such research is important to all educators who teach subjects that are quantitative. If experiments involving cooperative techniques show promise, then further research may prove fruitful.

This paper presents the results of test performances of two groups of first-year accounting students at a major Southeastern state university. All students received the same in-class lecture on two quantitative accounting topics by the same instructor. Approximately half the students were given a graded homework assignment to be completed by their group, while the second half had identical graded homework to be completed individually. Five to seven students were in each group, and all members of the group received identical grades. Later in the semester, the same students switched places. Those that had been given a group assignment received an individual homework assignment, and vice versa.

PRIOR STUDIES

A cooperative learning strategy allows students to work together on a graded assignment with the hope that group members will share knowledge within their group, thereby accomplishing a shared goal and increasing overall individual performance. An individual learning strategy requires students to work by themselves to accomplish their own goals (Johnson & Johnson, 1989). Encouraging students to work together has evolved from a grassroots effort by a few professors to an established method of education and learning. The goals of cooperative learning are diverse and include enhanced academic achievement and cognitive growth, increased student motivation, improved attitudes toward learning, social development and interpersonal relations (Natasi & Clements, 1991). Although all of these goals are important, the focus of this paper is restricted to the effect of cooperative learning techniques on individual academic achievement as measured by a common exam.

Some researchers have commented that merely placing students into groups and asking them to cooperate on a project will not be successful (Johnson & Johnson, 1990). These efforts often fail because student groups are afflicted with problems descriptively labeled as "free rider," "hitchhiker," "sucker," and "rich-get-richer" effects (Johnson & Johnson, 1990). Johnson & Johnson (1990) make the comment that " ... groups can also flounder through self-induced helplessness, diffusion of responsibility, social loafing, dysfunctional labor divisions, and destructive conflict."

While cooperative learning has positively influenced student performance and attitude in classroom settings (Sharon, 1980; Johnson & Johnson, 1989; Slavin, 1990), it has not always influenced performance when used with strategies originally designed for individual learning (e.g., graded homework assignments) (Carrier & Sales, 1987; Klein & Pridemore, 1992; Klein, et al., 1994). The above research suggests that a cooperative strategy may not affect educational outcomes in all settings. Therefore, the success of cooperative learning strategies is not assured, and its use may be more appropriate in some settings than others.

Related research suggests that an advantage of cooperative learning groups is that they give students an opportunity to talk aloud, challenge and defend a point of view, and focus on the problem-solving process rather than the answer (Gilbert-MacMillan, 1983). Parker (1984) found that small-group cooperative learning aids in developing thinking and problem-solving skills, and that this approach reduces student anxiety and competition by creating a friendly atmosphere, which allows students the freedom to learn from their mistakes. Another study of eighth-grade pre-algebra students found that students who worked cooperatively were better able to remember and apply problem-solving strategies than those students from independent practice classes (Duren & Cherrington, 1992).

The above findings lend credibility to the belief that cooperative learning techniques may increase individual learning when applied to quantitative accounting topics. In addition, this prior research suggests that cooperative learning can improve student attitudes toward the field of accounting.

HYPOTHESIS DEVELOPMENT

As described above, previous research has shown that cooperative learning can be effective in facilitating learning, particularly when dealing with quantitative topics in the field of mathematics. Therefore, it is reasonable to expect that cooperative learning techniques could enhance learning quantitative accounting topics as compared to using only the traditional lecture-recitation model and other methods that rely solely on individual efforts. However, it has not been established that cooperative learning techniques, specifically group work on graded homework assignments, are more effective than lecture-recitation and individually graded homework assignments in assisting students to learn quantitative rule-based accounting topics, such as inventory valuation and cost allocations. Effects previously identified (e.g., "free rider," "sucker," and "rich-get richer") may mitigate any gains from collaboration in a specific setting. In addition, although previous research has shown that cooperative learning does have positive recall and transfer effect, cooperative learning when applied to specific quantitative concepts may not transfer well to individual performance, which leads to the following hypothesis:

H1: There is a difference in individual test performance on quantitative questions between students completing graded homework assignments in cooperative groups, and students completing graded homework assignments individually.

Two specific quantitative accounting topics are investigated: cost allocations and inventory valuation. The hypothesis is non-directional since it has not been established whether the positive effects of cooperative learning are offset by negative effects in a specific accounting setting.

EXPERIMENT AND RESEARCH DESIGN

The subjects in this field experiment were sixty-nine students in two sections of an accounting principles course at a major Southeastern university. The format of the two sections was as similar as possible. Each section of approximately 35 students met with the same instructor each Tuesday and Thursday for 80 minutes throughout the semester. Section 1 met from 9:30--10:50 a.m., and Section 2 met from 11:00 a.m.--12:20 p.m. Students self-selected into groups of from five to seven students during the first week of the semester. The groups remained intact during the entire semester.

Switching the groups on the two graded assignments limited potential problems related to equivalency of subjects. Although the research design minimized the risk of problems, equivalency of the subjects was assessed because of its possible impact on interpretation of results. Data was collected on eight demographic variables: SEX, AGE, GPA, RACE, JOBHOURS per week, SEMHOURS (credit hours) enrolled during the current semester, declared MAJOR, and CREDITS (semester credit hours) earned prior to enrolling in the course. In addition, information was gathered regarding prior accounting coursework (COURSE), and any prior bookkeeping experience (EXP) of each student. Descriptive statistics and univariate tests for differences between the sections are shown in Tables 1 and 2, respectively. No significant differences between the groups were found (p-value = .05), although RACE was weakly significant (p-value < .10).

Of the overall sample of 69 students, 44 were male, 12 were planning to major in accounting, 11 were non-white, six had previous bookkeeping work experience, and 19 had previously taken accounting or bookkeeping coursework (generally in high school). The mean student was 21.7 years of age, had accumulated 57.7 previous credit hours, was currently enrolled in 14.9 credit hours, had a cumulative GPA of 2.97, and was working 11.2 hours per week.

Both sections were taught by one of the authors using a common syllabus. Class discussion and in-class exercises were the same for both sections. However the assignment of graded group versus graded individual homework assignments was reversed between the sections for two topics: cost allocations and inventory valuation. Each graded homework assignment was worth 5% of the total grade, and each member of a group received a common grade for the group assignment. The subjects then took a common multiple-choice examination, administered at a common time and place. The examination consisted of 50 multiple-choice questions, 13 of which were quantitative in nature. Of the 13 quantitative questions, three related to cost allocations, and three to inventory valuations. The remaining seven were general quantitative questions.

Univariate tests were used to assess the relative performance of the two subject sections on the test questions against the seven non-experimental sections relating to the specific quantitative topics and to assess the equivalency of the two subject sections. Data from all nine sections showed that students correctly completed 39.5%, 57.7% and 65.7% of the allocation, inventory valuation, and remaining questions, respectively.

Test performance of the two subject sections was not substantially different on the questions of interest from the other seven accounting sections not included in the experiment (allocations was .1% lower than the composite total, inventory valuation was 2.9% higher). There was no statistically significant difference in raw score test performance for the questions of interest between the two sections; however, section 2 marginally outperformed section 1 on the remaining test questions (p-value = 0.0757), and the test as a whole (p-value = 0.0995). Test performance was then regressed against the homework method used, test scores on other questions, and control variables to determine the significance and direction of the homework-method variable.

REGRESSION MODEL

The model as initially tested is as follows:

SCORE = + [[beta].sub.1]METHOD + [[beta].sub.2]MAJOR + [[beta].sub.3]EXP + [[beta].sub.4]SEX + [[beta].sub.5]RACE + [[beta].sub.6]COURSE + [[beta].sub.7]CREDITS + [[beta].sub.8]GPA + [[beta].sub.9]SEMHRS + [[beta].sub.10]AGE + [[beta].sub.11]JOBHR + [[beta].sub.12]QUEST

The variable of interest, METHOD, is a dichotomous indicator variable coded "0" for a group homework assignment, and "1" for an individual assignment. An additional variable, QUEST, is included and represents the non-quantitative test questions. It is included to allow modeling of the comparability of performance on the questions of interest and the remainder of the test. A description of all independent variables is included below as Table 3.

Correlation of the independent variables was examined. Most were not significantly correlated at the alpha = 0.05 level. Exceptions included the expected positive correlation of AGE with accumulated CREDITS, and negative correlation of JOBHRS with SEMHRS. In addition, MAJOR was significantly correlated with SEX (accounting majors tended to be female); CREDITS was significantly correlated with SEX (males tended to have accumulated more university credits by the time they took this course, possibly because as non-accounting majors they avoided accounting courses as long as possible); and AGE negatively correlated with SEMHRS (the few part-time students were older.) The highest correlation was 0.51 (SEMHRS with JOBHRS). Tests for multicollinearity for all regressions were performed. All variance inflation factors and condition numbers were well below the suggested values of 10 and 100, respectively, indicating that multicollinearity among these variables is not a problem. In addition, tests for heteroscedasticity, and analysis of residuals and autocorrelation, the Durbin-Watson D statistic revealed no violations of these assumptions. Analysis of the studentized residuals revealed no outliers that needed attention.

REGRESSION RESULTS

Test performance of the two subject sections was not substantially different on the questions of interest from the other seven accounting sections not included in the experiment (allocations was .1% lower than the composite total, inventory valuation was 2.9% higher). Data from all nine sections showed that students correctly completed 39.5%, 57.7% and 65.7% of the allocation, inventory valuation, and remaining questions, respectively. There was no statistically significant difference in raw score test performance for the questions of interest between the two subject sections; however, section 2 marginally outperformed section 1 on the remaining test questions (p-value = 0.0757) and the test as a whole (p-value = 0.0995).

Table 5 sets forth the coefficients, t-statistics and p-values of the ordinary least squares regressions on the full set of independent variables. The model is significant (F = 4.067, p-value = 0.001), and adjusted [R.sup.2] is 0.2220. The only significant independent variables are CREDITS and QUES, indicating that the number of semester credit hours accumulated prior to this course are positively associated with total test score, and that the non-quantitative questions on the test do have some correlation with the questions of interest. (Since the QUEST variable could be disguising common variance with other variables, a regression was run without QUEST. The variable GPA is then the only significant variable.) The variable of interest, METHOD, shows no indication of statistical significance (p-value = 0.92).

To derive a more parsimonious model, stepwise regression was performed with selection of the "best" model based on Mallows C(p) to minimize bias, and adjusted R2 to maximize explanatory power. The model developed after reduction by stepwise procedures is included as Table 6. The most appropriate parsimonious model for the dependent variable SCORE included SEX, CREDITS, AGE, RACE, and QUEST in the variable set, with only CREDITS and QUEST significant at the alpha = 0.05 level. The addition of METHOD provides virtually no change in the coefficients other than a 1% change in the value of the intercept. All the tests show clearly that METHOD has no statistically significant effect. Therefore the hypothesis is rejected.

To further understand the factors influencing test performance, a further regression was run to determine whether test scores as a whole were predictable. This regression of total SCORE on the full set of independent variables, excluding the METHOD variable, is also reported on Table 5. The model is significant (F = 4.547, p-value = 0.0001), and adjusted R2 is 0.3787. Although the predictive value has increased substantially for the test as a whole, only the GPA variable is significant.

CONCLUSIONS

The results of this field experiment indicate that test performance is not positively or negatively influenced by using the cooperative learning technique of graded group homework assignments versus graded individual homework assignments. This result is similar to previous mathematics studies where mathematics students exposed to cooperative learning situations learned as well as students in more traditional and individual-dependent learning strategies.

The most important finding of this study is that using the cooperative learning technique of graded group homework assignments versus graded individual assignments made no difference in individual test performance. Therefore accounting instructors may feel free to use this cooperative learning technique without fear that it may jeopardize learning quantitative accounting topics.

It is possible in the present study that there are positive effects of group learning but that they were mitigated by previously described negative effects (e.g., "free rider," "sucker," and "rich-get-richer"). If these effects could be controlled in a real-world setting, group assignments could lead to improved performance.

Performance on the quantitative questions of interest was not highly correlated with the non-quantitative questions. This lack of correlation may have been caused by a particular study strategy of the students, where students tend to study those topics that are easier to learn rather than the more difficult topics, such as cost allocations and inventory valuation. (Anecdotal evidence confirming this strategy was gathered during class discussions following the test.) It is not surprising that students found quantitative questions to be the most difficult to answer correctly.

The significance of the CREDITS variable in predicting quantitative test scores is somewhat unclear. However, it is very possible that students with more accumulated university credits have had more exposure to various quantitative topics. Therefore these students can more easily assimilate quantitative accounting topics than older students or those students with higher GPAs. Because of the significance of the CREDITS variable in predicting quantitative test scores, those who are involved in revising accounting and business curriculum may want to rethink where accounting principles courses are introduced to students. Students may benefit from being exposed to other quantitative courses before they are required to take accounting principles courses.

Although it was not tested here, the cooperative learning strategy employed by this study may have improved the overall attitudes of students towards fellow students, accounting, and business in general, as was found in several mathematics studies (Davidson, 1971; Olsen, 1973; Brechting & Hirsch, 1977; Chang, 1977; Shaughnessy, 1977; Treadway, 1983). Any positive change in students' attitude towards the field of accounting would be most welcomed by most business instructors, and is worthy of future study.

As with all research studies, there are many limitations. Care should be exercised in generalizing the results to other environments. The test subjects were all enrolled in a required introductory accounting course. While the sample represented a representative cross-section of predominately sophomore and junior business students at a large Southeastern university, they may not be representative of non-business students or students at other universities. Secondly, only those subjects enrolled in two sections instructed by one of the authors were included in the study. While all sections used a common textbook and methodology, and overall test scores appeared to be comparable between all other accounting sections, it is possible that results are not generalizable to other instructors. Thirdly, the specialized topics of cost allocations and inventory valuation were the topics of study. The effects of group versus individual study may vary for other topics. It is also possible that the cooperative learning technique chosen for this study did not have sufficient strength on its own to obtain either positive or negative results as measured by test performance.

There are ample opportunities to expand upon this research. Suggestions for future research include the following:

(1) Testing whether a student's opinion of the relative amount of learning group and individual assignments is positively correlated with actual performance.

(2) Testing whether a student's relative enjoyment of group versus individual assignments is correlated with relative learning.

(3) Testing whether a student's preference for the type of homework assignment is affected by either or both the student's belief regarding the relative amount of learning and the relative enjoyment.

(4) Testing whether a student's attitude toward the field of accounting is improved by the cooperative learning technique employed.

REFERENCES

Accounting Education Change Commission. (1990). Objectives for Accountants, Position Statement Number One, Bainbridge Island, WA: AECC

Accounting Education Change Commission. (1992). The First Course in Accounting, Position Statement Number Two, Bainbridge Island, WA: AECC

American Accounting Association. (AAA), Committee on the Future Structure, Content, and Scope of Accounting Education (The Bedford Committee). (Spring, 1986). Future accounting education: Preparing for the expanding profession. Issues in Accounting Education, 169-190.

Arthur Andersen & Co.; Arthur Young; Coopers & Lybrand; Deloitte, Haskins & Sells; Ernst & Whinney; Peat, Marwick, Main & Co.; Price Waterhouse; & Touche Ross. (1989). Perspectives on Education: Capabilities for Success in the Accounting Profession. New York, NY: Andersen et al.

Brechting, M.C. & C.R. Hirsch. (1977). The effects of small-group discovery learning on achievement and attitudes in calculus. American Mathematics Association of Two-Year Colleges Journal, 2, 77-82.

Carrier, C.A. & G.C. Sales. (1987). Pair versus individual work on the acquisition of concepts in a computer-based instructional lesson. Journal of Computer-Based Instruction. 14, 11-17.

Chang, P.T. (1977). On Relationships Among Academic Performance, Sex Difference and Attitude Persistence of Small Groups in Development College Level Mathematics Courses. Doctoral dissertation, Georgia State University, Dissertation Abstracts International, 38(7), 4002A.

Cooper, S., L. Cook, L. Smith, R. Mueck & J. Cuseo. (1990). Cooperative Learning and College Instruction: Effective Use of Student Learning Teams. Long Beach, CA: Institute of Teaching and Learning.

Davidson, N. (1971). The Small-Group Discovery Method of Mathematics Instruction as Applied to Calculus. Doctoral dissertation, University of Wisconsin, Dissertation Abstracts International, 31(11), 5927A.

Duren, P.E., & A. Cherrington. (February, 1992). The effects of cooperative group work versus independent practice on the learning of some problem-solving strategies. School Science and Mathematics, 80-83.

Gilbert-Macmillan, K.M. (1983). Mathematical Problem-Solving in Cooperative Small Groups and Whole Class Instruction. Doctoral dissertation, Stanford University. Dissertation Abstracts International, 4409A.

Johnson, D.W. & R.T. Johnson. (1989). Cooperation and Competition: Theory and Research. Edina, MN: Interaction Book Company.

Johnson, D.W. & R.T. Johnson. (1990). Cooperative learning and achievement. In S. Sharon (Ed.), Cooperative Learning: Theory and Research. New York, NY: Praeger, Inc.

Klein, J.D. & D.R. Pridemore. (1992). Effects of cooperative learning and need for affiliation on performance, time on task, and satisfaction. Educational Technology, Research and Development, 40.

Klein, J.D., J.A. Erchul & D.R. Pridemore. (1994). Effects of individual versus cooperative learning and type of reward on performance and continuing motivation. Contemporary Educational Psychology, 19, 24-32.

Natasi, B.K. & D.H. Clements. (1991). Research on cooperative learning: Implications for practice. School Psychology Review, 20(1), 120-131.

QOlsen, J.C. (1973). A Comparison of Two Methods of Teaching a Remedial Mathematics Course in Community College. Doctoral dissertation, Utah State University. Dissertation Abstracts International, 34(12), 7522A.

Parker, R. (1984). Small group cooperative learning in the classroom. OSSC Bulletin, 27(7), 56-60.

Sharon, S. (1980). Cooperative learning in small groups: Recent methods and effects on achievement, attitudes, and ethic relations. Review of Educational Research, 50, 241-272.

Shaughnessy, J.M. (1977). Misconceptions of probability: An experiment with a small-group, activity based, model building approach to introductory probability at the college level. Educational Strategies in Mathematics, 8(3), 295-316.

Slavin, R.E. (1990). Cooperative Learning: Theory, Research, and Practice. Englewood Cliffs, NJ: Prentice Hall.

Treadway, R.T. (1983). An Investigation of the Real Problem-Solving Curriculum in the College General Education Mathematics Course. Doctoral Dissertation, University of North Carolina at Greensboro. Dissertation Abstracts International, 45(1), 108A.

Douglass Cagwin, Lander University

Katherine J. Barker, Lander University

Table 1: Descriptive Statistics

Panel A--Dichotomous Variables

Demographic Statistic N # = 1 # = 0 % = 1 % = 0

MAJOR (1 = Accounting) 69 12 57 17.4% 82.6%
EXPerience (1 = Past Experience) 69 6 63 8.7% 91.3%
SEX ( 1 = Male) 69 44 25 63.8% 36.2%
RACE (1 = Non-white) 69 11 58 15.9% 84.1%
COURSE (1 = Previous Coursework) 69 19 50 27.5% 72.5%

Panel B--Continuous and Discrete Variables

Demographic Statistic N Mean Maximum Minimum

CREDITS 69 57.7 164 33
GPA 67 * 2.97 4.0 1.9
SEMHRS 69 14.9 18 3
AGE 69 21.7 43 19
JOBHR (per week) 69 11.2 45 0

* Two subjects transferred from another institution at the
beginning of the semester and had no accumulated GPA.

Table 2: Univariate Tests of Group Equivalency (a)

Demographic Statistic Mean

Sec 1 Sec 2 Std Dev
Sec 1 Sec 2 Parametricb
Statistic P-value Non-parametric
Statistic P-value

MAJOR (1=Accounting) 0.18 0.17 0.39 0.38 0.05
EXPerience (1=Past Exp) 0.12 0.06 0.33 0.24 0.88 (c)
SEX (1=Male) 0.71 0.57 0.46 0.50 1.16
RACE (1=Non-white) 0.24 0.09 0.43 0.28 1.70 (c)
COURSE (1=Previous 0.32 0.23 0.47 0.43 0.88 (c)
coursework)

CREDITS 58.0 57.5 25.0 15.2 0.1
GPA 3.0 3.00 0.60 0.60 -0.25
SEMHRS 15.4 14.5 1.78 3.20 1.49 (c)
AGE 20.9 21.6 2.34 5.20 -0.74 (c)
JOBHR (per week) 9.9 12.4 12.4 14.3 -0.76

Demographic Statistic Mean

Sec 1 Sec 2 Std Dev
Sec 1 Sec 2 Parametricb
Statistic P-value Non-parametric
Statistic P-value

MAJOR (1=Accounting) 0.96 0.003 >0.25
EXPerience (1=Past Exp) 0.39 0.79 (d) >0.25
SEX (1=Male) 0.25 1.35 (d) >0.25
RACE (1=Non-white) 0.01 2.88 (d) 10>p>.05
COURSE (1=Previous 0.39 0.95 (d) >.25
coursework)

CREDITS 0.92 0.81 (e) 0.42
GPA 0.8 -0.09 (e) 0.93
SEMHRS 0.15 1.34 (e) 0.18
AGE 0.46 0.08 (e) 0.93
JOBHR (per week) 0.45 -0.49 (e) 0.63

(a) N = 34 for Section 1 and N = 35 for Section 2; except for GPA,
which has N = 33 and N = 34, respectively.

(b) T-tests of differences between means.

(c) Failed F-test that variances are equal; results computed with
Cochran Procedure.

(d) Test of equal proportions Chi-Square statistic.

(e) Wilcoxon Rank Sum Test Z-statistic.

Table 3: Independent Variable Descriptions

Variable Name Variable Description

METHOD Indicator variable where 1 = group homework
 assignment, and 0 = individual homework assignment.

MAJOR Anticipated major field of study, indicator variable
 where 1 = Accounting; 0 = Not
 Accounting (other business major).

EXP Indicator variable where 1 = previous accounting or
 bookkeeping work experience, 0 = no previous
 experience.

SEX Indicator variable where 1 = Male, 0 = Female.

RACE Indicator variable where 1 = Non-white, 0 = White.

COURSE Indicator variable where 1 = previous accounting or
 bookkeeping coursework, 0 = no previous coursework.

CREDITS Number of college credit hours completed prior to
 current semester.

GPA Current grade-point average on a four-point scale.

SEMHRS Number of credit hours enrolled in for current
 semester.

AGE Student age in years at time of examination.

JOBHR Number of hours per week of employment.

QUEST Percentage of correct questions on the examination
 that were on topics other than the quantitative
 questions of interest: cost allocations or inventory
 valuation.

Table 4: Correlation Matrix of the Independent Variables

Variable MAJOR EXP SEX CREDITS GPA

MAJOR 1.00
EXP .13 1.00
SEX -.37 -.20 1.00
CREDITS -.22 .16 .27 1.00
GPA .16 .03 -.17 -.12 1.00
SEMHRS -.01 -.06 -.18 -.23 .22
AGE .05 .03 .08 .41 .07
RACE .22 .01 -.17 -.13 -.11
COURSE .23 .16 -.21 -.11 .03
JOBHR -.16 .08 .16 .25 -.23

Variable SEMHRS AGE RACE COURSE JOBHR

MAJOR
EXP
SEX
CREDITS
GPA
SEMHRS 1.00
AGE -.33 1.00
RACE .19 -.12 1.00
COURSE -.02 .08 .00 1.00
JOBHR -.51 .18 -.20 .17 1.00

Note: Bold type = significant at alpha = 0.05 level.

Table 5: Results from OLS Regressions of SCORE and TOTAL on
Independent Variables

 SCORE (b)

F-statistic = 4.067 [R.sup.2] 0.2943
p-value = 0.0001 Adj [R.sup.2] 0.2220

Independent
Variable Coefficient t-statistic p-value

Intercept -55.76 -1.58 0.11
MET HOD -0.51 -0.10 0.92
MA JOR 4.01 0.51 0.61
EXP 1.09 0.11 0.91
SEX -6.01 -0.97 0.33
RACE -12.92 -1.70 0.09
COU RSE -2.18 -0.34 0.73
CRE DITS 0.33 2.12 0.04
GPA 4.67 0.78 0.44
SEM HRS -0.09 -0.06 0.95
AGE 0.86 1.11 0.27
JOB HR 0.02 0.07 0.10
QUEST 1.84 3.02 0.00

 TOTAL (b)

 F-statistic = 4.547 [R.sup.2] 0.4855
 p-value = 0.0001 Adj [R.sup.2] 0.3787

Independent
Variable Coefficient t-statistic p-value

Intercept 22.16 1.28 0.21
MET HOD N/A N/A N/A
MA JOR 2.74 0.72 0.47
EXP -0.82 -0.17 0.86
SEX -2.30 -0.77 0.44
RACE -5.58 -1.52 0.13
COU RSE -4.16 -1.37 0.18
CRE DITS 0.11 1.39 0.17
GPA 12.98 4.64 0.00
SEM HRS -0.41 -0.58 0.56
AGE 0.61 1.51 0.14
JOB HR -0.08 -0.66 0.51
QUEST N/A N/A N/A

(a) where SCORE is the prediction of the quantitative questions.

(b) where TOTAL is the prediction of the total test scores.

Table 6: Results for OLS Regressions of SCORE on Reduced Set
of Independent Variables

Without METHOD

F-statistic = 10.203
p-value = 0.0001
[R.sup.2] 0.2850
Adj [R.sup.2] 0.2570

Independent Coefficient t-statistic p-value
Variable

Intercept -52.05 -2.78 0.01
SEX -7.05 -1.29 0.20
CREDITS 0.32 2.25 0.03
AGE 0.89 1.27 0.21
RACE -12.53 -1.77 0.08
QUEST 2.15 4.70 0.00
METHOD

With METHOD

F-statistic = 8.438
p-value = 0.0001
[R.sup.2] 0.2850
Adj [R.sup.2] 0.2513

Independent
Variable Coefficient t-statistic p-value

Intercept -51.81 -2.73 0.01
SEX -7.05 -1.28 0.20
CREDITS 0.32 2.24 0.03
AGE 0.88 1.26 0.21
RACE -12.53 -1.76 0.08
QUEST 2.15 4.68 0.00
METHOD -0.50 -0.10 0.92