Do Peers Influence Achievement in High School Economics? Evidence from Georgia's Economics End of Course Test

Abstract

The authors provide the first estimates of the impact of peers on achievement in high school economics. The estimates are obtained by analyzing three years of data on all high school students who take Georgia's required economics course and its accompanying high-stakes End of Course Test (Georgia Department of Education). They use an instrumental variables approach with teacher-level fixed effects to control for selection bias, simultaneity, measurement error in the measure of peer quality, and nonrandom assignment of teachers to students. The authors find that an increase of one standard deviation in the prior academic achievement of peers increases achievement in economics by 0.03 standard deviation.

Keywords:

• high school economics
• instrumental variables
• peer effects

JEL Codes:

• A21
• C30
• I21

This research was sponsored by a grant from the Council for Economic Education through funding from the U.S. Department of Education Office of Innovation and Improvement. The authors thank the Georgia Council on Economic Education and Georgia College and State University for matching funding. They also thank the Georgia Department of Education for providing the data. Finally, they thank the editor and two anonymous referees for comments and suggestions.

Notes

1. The EOCT (GaDOE 2008) is a standardized test of a standardized state curriculum. Every economics class in Georgia is expected to cover the domain areas on the EOCT.

2. Levi (1973, 985) stated, “if we have only one independent variable that we expect to have measurement error among other independent variables that are measured without error, then in large samples: (i) The direction of the bias on the coefficient of the variable measured with error is unambiguously downwards, despite the inclusion of the other independent variable measured without error.” We feel our independent variables (other than PEER_ECON) are measured without error. Therefore, our OLS estimates of the magnitude of peer effects are biased toward zero because of measurement error. See, for example, Greene (2000) or Wooldridge (2006) for further explanations of why this is the case.

3. A concern with the data used by Gaviria and Raphael (2001) is that their school peer behavior variables are constructed from only 13.3 students per school, on average.

4. We estimated the Hansen's J overidentification test, which is designed to test whether the instruments in an IV model are exogenous. Unfortunately, the Chi-square test statistic of Hansen's J overidentification test is large enough to reject the null hypothesis that the instruments are exogenous. However, since the J statistic is essentially the sample size times the minimized value of the objective function, our large sample size may be to blame for the rejection. We thank an anonymous referee for this insight.

5. The required EOCT exams are given in the following subjects: algebra, geometry, U.S. history, economics/business/free enterprise, biology, physical science, ninth-grade literature and composition, and American literature and composition (GaDOE 2008).

6. The EOCTs are officially administered in the spring and the fall. Some students take remedial EOCTs in the summer. These tests are not included in the data.

7. Of the 90 questions, 75 count toward the student's score. Each test also field tests 15 questions that do not count toward the student's score.