Economic education, experimental methods and the structure-conduct-performance paradigm.

Ray, Margaret A.

Introduction

The traditional view of student learning in the economic education literature places the teacher and/or school in the role of producer while students are seen as inputs to which value is added. Empirical models of economic education often assume students are passive receptors, with varying degrees of initial endowments, subject to the teaching techniques and learning environment supplied by teachers and schools. Within this framework, teaching effectiveness, and thus student learning, can be changed by simply altering the teaching technology or school environment. However, most educators would agree that in most circumstances the ultimate control over teaching effectiveness lies with the student, and not with the "teacher/school producer."

Consider an alternative economic model of education that views students as the producers of learning. Imagine the classroom as a market, a market in which students produce learning. In this market, the output is monitored by test scores. The student "produces" test scores by learning and trades the scores for grades. Instructors strive to assure that grades are highly correlated with actual learning. The economic goal of students is to maximize learning (as measured by grades) subject to a time constraint (the student's available time and the length of the course) or, alternatively, to minimize study time subject to a given level of learning (grades).

This alternative view is consistent with research findings of sociologists studying the motives and goals of college students. Empirical research has established the "grade point average perspective" as prevalent among college students:

The student's generalized goal enjoins him to be serious about college: to recognize it as a serious place where important things happen and to try to do well in all areas of college life as a sign of achieved maturity. His perspective on academic work develops as he interacts with other students in an environment in which, as we shall see, grades are the chief form of institutionalized value and the institutional basis of punishment and reward in academic pursuits (Becker, Geer and Hughes, 1983).

Becker, Geer and Hughes further recognize the "analogy" grade directed behavior and behavior within a money economy. Students strive to make grades which are "good enough" to pass their individual quality standards just as workers strive to earn income to maintain a desired standard of living. (Obviously the desired standards and the ability to achieve those standards vary across individuals in both cases.) Faculty often disagree with the emphasis placed on scores, grades, and grade point averages but the institutional structure of universities has a deeply embedded "grade point perspective."

The instructor may be seen as serving three functions in the market. First, the instructor "buys" learning with grades. (In most circumstances the instructor is the sole buyer of test scores--a monopsonist with a degree of control over the "price" of grades). A second function of the instructor is as an input. Classroom instruction is combined with readings, homework, study time, student ability and other inputs to produce learning. A quality instructor increases learning efficiency and thereby decreases the cost of student learning, while a poor instructor increases the cost. The last, and perhaps most important, function of the instructor is to determine the operational structure of the course. The course structure creates the market which correlates grades with learning.

This alternative perspective places students in the role of producers. The student makes decisions that determine if and how much learning will be produced. Within this framework, students, rather than teachers, are endowed with the power and responsibility to produce learning. Teachers have no direct control over student actions, they can only attempt to induce learning by providing the optimal class structure and teaching inputs. Adopting this classroom market perspective of the education process has several benefits. One is the ability to borrow the "structure-conduct-performance" paradigm from the field of industrial organization.

In industrial organization, market structure represents the firm's external environment. When the classroom is viewed as a market, structure refers to the environment set up by the teacher. Each student must operate within the confines of the class structure.

Firm behavior concerns the conduct of the firm within the market structure. Student behavior refers to the choices made by students within the class structure. The link between structure and performance is often overlooked or ignored in the "teacher/school producer" models of learning. Teachers have no direct control over student behavior or performance. Performance can only be affected via the teachers ability to influence student behavior using class structure.

Finally, market performance is evaluated in terms of efficiency, equity, and progress. In education, academic performance can be evaluated using testing instruments. Grades, student evaluations, and scores on standardized exams are all measures of academic performance.

How may the classroom market perspective be useful for research in education? Importantly, it calls attention to alternative class structures, student behavior and academic performance within the structure-conduct-performance paradigm. This framework incorporates the interrelationships between the class structure teachers create, the students' behavior choice and academic performance.

An additional benefit of this alternative approach derives from the ability to use experimental methods to study market organization. The ability to create markets and study them using experiments gives researchers an additional method of investigation (Taylor, 1988). Since the classroom is viewed as a market, this approach suggests that educational issues can be investigated using experimental methods similar to those used in the field of industrial organization. For example, it is possible to investigate how alternative class structures affect student performance using this method. Questions of how many tests should be given, should homework be assigned, do quizzes increase learning and so forth can be addressed using experimental methods.

Experimental Methods in Education: An Example

Experimental methods are now used by economists in several fields. Economic experiments often use university students as the economic agents operating in a market designed by the experimenter. Analyzing student behavior in these strategic situations gives insight into the behavior of economic agents in the type of market under investigation. At the same time, these experiments give insight into student behavior. For experiments using students to shed light on firm behavior, it must be assumed that students in the experiment behave as firms behave in a market. But for experiments to shed light on student behavior, it must only be assumed that students behave as students.

In the university classroom, students face interdependence. The outcome of a student's decision often depends on other students. For example, the pace of the class may be determined by the top students, the slower students, or the median student; the grade scale may be set based on the highest score or the mean score; and teacher expectations, which determine the "terms of trade" between scores and grades, may be affected by the abilities of class members as a whole. Experimental methods have been used to study firm behavior in interdependent markets. These same methods can be applied to study student behavior in the classroom. Specifically, experiments can be designed to determine what factors affect student conduct and performance when they are placed in a "strategic" classroom situation.

The market in which students operate in a classroom experiment is a familiar market with which they have much experience. Below is one example of how experimental methods similar to those used to investigate questions in industrial organization can be applied to economic education.

The Experiment

The experiment was conducted at Texas Christian University (TCU). Participants were students enrolled in Macroeconomic Principles during the Fall 1989 semester. In this experiment, students chose whether to take an optional weekly quiz. If the choice were between taking a quiz with some probability of earning bonus points at zero cost and not taking a quiz to earn no bonus points there is no real economic choice. Therefore, in this experiment, a student could earn bonus points either by taking the quiz or by not taking the quiz. Students are placed in a strategic situation where the return to their decision depends on the decision of other students in the class. Student instructions and payoffs are included as Appendix 1. If a student chose to take the quiz she would receive either her score (if the class chose not to take the quiz) or the class average (if the class chose to take the quiz). If she chose not to take the quiz she would receive either a 0 (if the class chose not to take the quiz) or 100% (if the class chose to take the quiz). (As reported in the Appendix, the "class decision" to take the quiz required an 80% participation rate.) The payoffs are similar to the payoff in a "prisoner's dilemma" game for those choosing not to take the quiz. Institutional constraints, however, mandated that students receive some credit if they chose to take the quiz.

Quizzes were administered during each Friday class over the course of the 15 week semester. Students were first given a choice sheet to mark their choice (to take the quiz or not to take the quiz). Their response was concealed and collected. After collecting all responses, quizzes were distributed.

Each Monday students were given the results of the Friday quiz. The results included both the student's individual and class choices, individual quiz score (when applicable), individual points earned for the current round and total points earned. Also reported were the class quiz average and the exact percent of the class that chose to take the quiz. At the end of the semester, bonus quiz points were added to final class grades at a rate of 15 points = 1%. That is, a student earning perfect scores on quizzes (150 total points) would earn a 10% bonus.

The Model

This experiment may help instructors understand student behavior by observing students in a situation similar to that faced by interdependent firms in a collusive agreement. Students chose whether to take the optional quiz. They could either study and earn points by taking the optional quiz or they could try to earn points, with zero study costs, by choosing not to take the quiz (thus "free-ride" on the class decision). Note from the payoffs outlined in Appendix 1 that the return to the student's individual decision depends on the decision of the class as a whole. The adopted 80% decision rule for the class required approximately 40 students to take the quiz (assuming perfect class attendance). Eighty percent was chosen for the decision rule to create a reasonable probability that students could indeed earn points by choosing not to take the quiz. The choice was somewhat arbitrary, relying on the instructors ability to approximate the students' perceptions of probabilities. Changing this percentage would alter the student's perceived probabilities and therefore would be expected to alter the results.

Compare the students situation to the situation of the firm. The firm may avoid excessive competition and maximize joint profits by colluding. The class may avoid competition and maximize joint bonus points by taking the quiz. In effect, the class group may be viewed as entering a collusive agreement in which each student studies and takes the quiz. But in each case, the possibility of free-riding exists. A firm may cheat on the agreement in hope of earning high profits. A student can violate the implied agreement by choosing not to take the quiz, in hope of earning a score of 100%. In either case, the action would be successful only if other firms/students do not also decide to free-ride. Putting students in this strategic situation, makes it is possible to observe behavior when choices are interdependent, and to determine what variables influence a students decision to study.

The Results

The percent of the class which chose to take the optional quiz during each time period varied. In only four of fourteen periods (weeks) did 80% or more students choose to take the quiz. The highest percent choosing to take the quiz during the semester was 83%. It is also interesting to note that each student attempted, at least once, to earn the full 10 points by not taking the quiz. This indicates that the variation in the percentage of the class taking the quiz represents decisions by all students, not just a small segment of the class. The average quiz score ranged from 3 to 7 points, perhaps indicating the "bonus" nature of the quizzes.(1)

In making the decision each week, students can consider information gained from the results of all previous periods. In addition, students know their own grades (including midterm exams, homework), how much they have studied, and how well they understand the material. Using this information, students can decide which choice yields the highest expected return. The expected return derives from the student's perception of the probabilities of the outcomes and the perceived return from each outcome

The factors affecting the decision of the class to study for the quiz give insight into student conduct. To investigate which information significantly affected student decisions, the following relationship was tested using a simple regression framework:

Percent = A0 + B1 Average + B2 Score + B3 Midterm + B4 Attendance + e

Where;

Percent = % of class in attendance choosing to take weekly quiz

A0 = OLS regression constant

Average = Cumulative class average the day of each quiz

Score = Class average on previous week's quiz

Midterm = 0 before midterm exam; = 1 after midterm exam

Attendance = Class attendance on day of weekly quiz

e = Random error term

The cumulative class Average represents the average student grade at the time of the quiz (including homework and midterm grades). A low class average may encourage students to take the quiz in order to earn much needed points or may cause students to choose not to take the quiz if they feel "free-riding" behavior is their only hope.

The Score variable measures the class average on the previous quiz. Class success in taking the last quiz may influence student decisions if it changes student expectations for success on the current quiz. Success on the last quiz will lead to more students taking the current quiz if the expected return to taking the current quiz increases.

The midterm grade is included to measure changes in student behavior that occur after the midterm exam. The class average for the midterm exam was 76%. High midterm exam grades will affect student decisions if expected returns are changed. For example, sufficiently high midterm grades may cause students to become satisfied with current grades so that additional studying is not deemed necessary.

The percent attending class determines the number of students that may earn points by not taking the quiz. The Attendance variable may affect student perceptions of the probability of receiving points by not taking the quiz.

Not included in this equation are factors that are outside of the instructor's control and are not easily measured. For example, a party the night before the quiz or an exam given in another class can also be expected to enter into a student's decision.

The results of the estimation are included in Table 1. These results indicate short run success (represented by the Score coefficient) significantly increased the percent of the class which takes the quiz. However, in the long run student success (as measured by the Average coefficient) significantly decreased the percent of students taking the quiz. This result implies that students may increase or decrease studying as a result of success on recent assignments. However, reaching a high or low cumulative class average will discourage or encourage studying respectively. It appears that students will work to obtain the desired final grade, but short-run decisions are affected by recent success on assignments and exams.

The Midterm coefficient is statistically significant and conforms with the hypothesis that student success in the long-run (i.e., achieving the desired grade) may decrease the percentage of students choosing to take quizzes.

These results are consistent with the findings of Becker, Geer and Hughes.

TABLE 1
REGRESSION RESULTS
(Dependent Variable = Percent)

Variable Coefficient t-Ratio

Constant 357.32 3.30(***)
Average -5.37 2.89(**)
Score 0.20 1.92(*)
Midterm -21.47 1.96(*)
Attendance -0.02 0.06

F = 8.45 |R.sup.2~ = 0.57 |R.sup.2~ = 0.56 N = 15

*** Significant at the .01 level, two tailed test
** Significant at the .05 level, two-tailed test
* Significant at the .10 level, two-tailed test

A certain minimum is essential because of the way the valuable |grade~ is institutionalized. But beyond the minimum, which represents the level necessary in order to have the privilege of choosing where to put one's remaining time and effort, choice becomes possible and the person finds that he must balance the various rewards available against one another in making that choice. (Becker, Geer and Everett, 1983).

The idea, of course, is that as students reach the minimum desired grade, the opportunity cost of studying becomes very high.

Conclusions

Educators are aware of the futility of trying to teach students who are not interested in learning. Regardless of the amount of time, talent and energy an instructor expends, the student has ultimate control over success in the classroom. Viewing the classroom as a market, where students combine educational inputs to produce grades allows us to study the education process using the structure-conduct-performance paradigm. This paradigm is useful for educators and researchers interested in determining appropriate class structures, establishing particular student conduct, or improving academic performance.

Experimental methods have been applied to areas throughout the economics discipline. They have been especially useful for studying topics in industrial organization. In addition to providing a framework for investigating the production of learning, use of the Structure-conduct-performance paradigm invites the use of experimental methods for the study of education. The experiment detailed here is merely an example of the type of research that can be undertaken if classroom behavior is treated as a market. The importance of this paper lies in the ideas and research that may result from the adoption of this alternative perspective and experimental methodology to analyze student learning.

References

Becker, Howard, Blanche Geer and Everett Hughes, "Making the Grade" in Howard Robby and Candace Clark, eds. Social Interaction. New York: St. Martins Press, 1983.

Taylor, Herb, "Experimental Economics: Putting Markets under the Microscope." Philadelphia Federal Reserve Business revue, March/April, 1988.

Notes

1. The percentage of the class that choose to take the quiz and the mean quiz score for each week were as follows:

Week 1 (41%, 3), Week 2 (82%, 7), Week 3 (41%, 5), Week 4 (62%, 5), Week 6 (83%, 3), Week 7 (70%, 4), Week 8 (62%, 5), Week 9 (66%, 6), Week 10 (64%, 3), Week 11 (80%, 4), Week 12 (80%, 4), Week 13 (76%, 6), Week 14 (72%, 5)

Appendix 1

Economics 2103 10:00 MWF

Bonus Quiz--Instructions

You have the opportunity to participate in an experiment involving the allocation of your scarce economic resources. You may choose not to participate if you wish, however the results can not adversely affect your grade while they may increase it. To participate you will need to purchase scantron answer forms from the bookstore.

The Experiment

You will have the chance to practice with two quizzes prior to beginning the experiment. There will be 15 rounds (quizzes) during the semester. To participate, you must choose whether or not to take the quiz being given. You may earn quiz points either by choosing to take the quiz or by choosing not to take the quiz. If you choose not to participate or are not in class when the quiz is given, you will not earn quiz points.

The Payoff

The figure below shows the quiz points earned in each possible situation, given your choice. Your return depends on whether or not "the class" has chosen to take the quiz. "The class" is considered as having taken the quiz if 80% or more of those present choose to take the quiz. Each quiz is worth 10 points.

The results (your score, your quiz points, the % taking the quiz, the class average) will be reported for each quiz prior to the next round.

Margaret A. Ray is an Assistant Professor of Economics, Texas Christian University, Fort Worth. The author wishes to thank Paul W. Grimes, Joe K. Law and Paul S. Nelson for their comments and suggestions. Any remaining errors are the sole responsibility of the author.