Is class attendance a proxy variable for student motivation in economics classes? An empirical analysis.

Durden, Garey ; Ellis, Larry

Introduction

In recent studies, Romer (1993) and Durden and Ellis (1995; 1998) have shown that class attendance appears to affect student average scores and resulting grades in economics classes. Specifically, the fewer absences a student has, the higher the overall exam average and the greater the probability that a given student will earn an A or B. In his study, Romer also attempts to answer the question, Do class attendance rates and internal motivation separately influence performance, or is attendance simply a proxy for the motivation? His results suggest that both factors affect how well a student does, requiring that the influence of each be accounted for.

This study provides an additional test of the relative effects of absence rates and of the effects of internal student motivation. This analysis, however, differs from that of Romer in several ways. Whereas Romer's equations are based on samples ranging from 116 to 195 students in an intermediate macroeconomics class taught by a single individual at a research institution, this study is based on a sample of 252 students enrolled in a principles of economics course taught by two different professors at a comprehensive university. Moreover, Romer's data on absence rates were derived from attendance records limited to six class meetings where he took roll; this study is based on absence rates for a full semester. Finally, while Romer's model is minimally specified, including only absence rates, prior GPA, and his proxy for motivation (the fraction of assigned problem sets completed by each student), this model is more fully realized. Such differences will provide a useful refutation or corroboration of Romer's work.

The Influence of Student Motivation and Other Factors

The dependent variable used for this study is the overall class average for each student. Since there are two instructors represented, observations are normalized so that the mean score for each is equivalent. The independent variables (1) were constructed from responses to a survey administered to students during the fall semester of 1996. These variables include the number of Absences from class, GPA, SAT Math score, SAT Verbal score, and Semester Hours Completed, which are all continuous variables. GPA and SAT scores were independently verified by someone qualified to review relevant materials. This study also contains three dummy variables: Male, with value=1 if male, 0 if female; Calculus, with value=1 if the student has taken calculus, 0 otherwise; and Job, with value=1 if the student is working, 0 otherwise.

In order to measure the influence of motivation, the authors of this study constructed two variables. The first variable is a composite, which is based on answers to several motivation-related questions from the survey. These questions seek to determine why the student enrolled in the course, what is the perceived difficulty of the course compared to others taken by the student, whether the student had a personal interest in the course content, if the student would take another economics course, and whether the respondent was motivated to do well in economics, relative to other courses. On each question, the intensity ranged from 1, low motivation, to 3, the highest motivation level. Using the information obtained, the authors of this study created Motive 1, which, for each respondent, is the numerical sum of answers to the motivation-related questions. Values for Motive 1 range from 5 to 14. On an additional survey question, students were asked directly, "On a scale of 1 to 3 (with 1 being lowest and 3 being highest) how would you rank your motivation to do well in this course?" Responses to this question constitute the variable, Motive 2, with range from l to 3.

Empirical Results

In the regressions that follow, Average Score for each respondent is the dependent variable. The mean and standard deviation for the dependent and independent variables along with d-scores for the coefficients in the regressions cited below are reported in Table 1.

The first OLS regression, containing no measure of motivation, yielded the following results, with all coefficients significant at 5% or better except Job and SAT Verbal:

25.119 - .4429 * (Absences) + 11.72 * (GPA) + .0137 * (SAT Math) +
(6.56) (2.40) (13.21) (2.13)
.0099 * (SAT Verbal) +
(1.53)

2.82 * (Male) + 3.17 * (Calculus) + .0380 * (Semester Hours Completed)
(3.31) (3.17) (2.07)
+ .1322 * (Job)
(0.16)

[R.sup.2]=.573, N=252

(t-values in parentheses)

When Motive 1 is entered, the following results were obtained:

12.718 - .4233 *(Absences) + 11.73 * (GPA) + .0127 * (SAT Math) +
(3.15) (2.46) (14.21) (2.12)
.0093 * (SAT Verbal) +
'(1.54)

1.84 * (Male) + 2.25 * (Calculus) + .0436 * (Semester Hours Completed)
(2.27) (3.06) (2.53)
- .4793 * (Job) +
 (0.61)

1.61 * (Motive 1)
(6.54)

[R.sup.2]=.637, N=252

(t-values in parentheses)

The Motive 1 variable is highly significant, and its entry increases R-Squared from .573 to .637. Between the two regressions, only minimal changes occur with respect to coefficient sizes (and t-values) for Absences, GPA, SAT Math, and SAT Verbal. This suggests that these variables do not serve as proxy variables for motivation. Among this sample of students, motivation appears to be an independent influence. If so, classroom performance models that do not control for motivation will be misspecified.

When the direct motivation variable, Motive 2, is entered, this study yields the following results:

19.924 - .3297 * (Absences) + 11.19 * (GPA) + .0134 * (SAT Math) +
(4.90) (1.79) (12.67) (2.11)
.011 * (SAT Verbal) +
(1.75)

3.20 * (Male) + 2.70 * (Calculus) + .0357 * (Semester Hours Completed)
(3.79) (2.27) (1.97)
+ .2440 * (Job) +
(0.29)

2.53 * (Motive 2)
(3.33)

[R.sup.2]=.592, N=252

(t-values in parentheses)

For this version, the R-Squared change is smaller, and the coefficient on Absences is somewhat reduced, thus suggesting that class attendance and personal motivation are related. If this is true, when motivation is not independently controlled for, the effect of absence from class will be overstated.

The d-scores reported in Table 1 for the coefficients in each of the three regressions are computed as the t-value on the coefficient divided by the square root of N. As such, they provide an effect size measure for evaluating the relative importance of the independent variables. (2) In the case of the motivation variables, for example, the d-scores indicate that the Motive 1 variable is relatively more important and has a larger effect on student performance than the Motive 2 variable. It remains uncertain, however, why Motive 1 seems to play a more important role than Motive 2 in affecting student performance. Since it is an indirect, composite measure drawn from several questions, perhaps it does a better job of capturing the student's desire to do well in the course.

Summary and Conclusions

In his 1993 paper, Romer suggests that, in classroom performance models, student motivation to do well should be controlled for. Otherwise, the effect of attendance and other possible motivation-related factors may be overstated, and the predictive value of empirical results may be compromised. Using data from students in intermediate macroeconomics, he found that attendance and student motivation were both significant determinants of average exam scores.

This paper uses data from classes in principles of economics and a more fully specified model than that employed by Romer to address the question, "Is personal motivation an independent determinant of classroom performance, or do variables such as class attendance, GPA and SAT scores act as proxy measures for motivation?" The results suggest, rather strongly, that motivation is an independent factor with regard to average scores earned. There is some evidence that classroom attendance may serve as a proxy for the effects of internal motivation, but the effect is rather weak. In any case, the evidence presented here confirms Romer's idea that motivation must be specifically controlled for if empirical models are to generate reliable estimations and predictions.

Table 1
Characteristics of Variables

 d-scores
 Standard
Variable Mean Deviation Regression (1) (2) (3) *

Average Score 72.969 9.922 -- -- --
Absences 2.688 2.328 .151 .155 .113
GPA 2.772 0.525 .832 .895 .798
SAT Math 514.137 73.013 .134 .134 .133
SAT Verbal 458.404 72.457 .096 .097 .110
Male 0.514 0.501 .209 .143 .239
Calculus 0.538 0.499 .199 .195 .143
Semester Hours
Completed 49.780 24.430 .130 .159 .124
Job 0.441 0.497 .010 .038 .018
Motive 1 8.584 1.690 -- .412 --
Motive 2 2.392 0.590 -- -- .210

* The d-scores are for each of the coefficients of the indicated
variable in the three regressions reported. Regression (1), (2)
and (3) indicate the regressions in the order in which they appear
in the paper.

ENDNOTES

(1) Selection of our independent variable set is based on convention established in earlier work. See Durden and Ellis (1995) for a survey of these studies.

(2) Conventional guidelines for evaluating the relative importance of a variable are d=.20 is a small effect; d=.50 is a medium effect; and d=.80 is a strong effect. See Cohen (1988).

REFERENCES

Brasfield, David, James McCoy, and Martin Milkman. "The Effect of University Math on Student Performance in Principles of Economics." Journal of Research and Development in Education 25 (Summer 1992):240-47.

Cohen, Jacob. Statistical Power Analysis for the Behavioral Sciences. Hillsdale, NJ: Lawrence Erlbaum Associates, Publishers, 1988.

Durden, Garey C., and Larry V. Ellis. "The Effects of Attendance on Student Learning in Principles of Economics." American Economic Review 85 (May 1995):343-46.

Lumsden, Keith G., and Alex Scott. "The Economics Student Reexamined: Male-Female Differences in Comprehension." Journal of Economic Education 18 (Fall 1987):365-75.

Park, Kang H., and Peter M. Kerr. "Determinants of Academic Performance: A Multinomial Logit Approach." Journal of Economic Education 21 (Spring 1990): 101-11.

Romer, David. "Do Students Go to Class? Should They?" Journal of Economic Perspectives 7 (Summer 1993): 167-74.

Siegfried, John J., and Rendigs Fels. "Research on Teaching College Economics: A Survey." Journal of Economic Literature 17 (September 1979):923-69.

Williams, Mary L., Charles Waldauer, and Vijaya G. Duggal. "Gender Differences in Economic Knowledge: An Extension of the Analysis." Journal of Economic Education 23 (Summer 1992):219-31.

GAREY DURDEN and LARRY ELLIS are Professors of Economics at Appalachian State University in Boone, North Carolina.