Quasi-Experimental Evidence of Peer Effects in First-Year Economics Courses at a Chinese University

Abstract

The authors of this article implement a quasi-experimental strategy to estimate peer effects in economic education by exploiting the institutional setting in a large public university in China, where roommates are randomly assigned conditional on a student's major and province of origin. They found significant impacts of peer academic quality, measured as roommates’ average scores on the national College Entrance Exam, on first-year economics students’ scores in first-year microeconomics, macroeconomics, and accounting courses. They also found nonlinearity in peer effects: Roommates’ academic ability has significant effects for academically weak students but not for academically strong students.

Keywords:

• economic education
• peer effects
• quasi-experiment

JEL codes:

• C22
• I21

ACKNOWLEDGEMENTS

The authors thank JEE editor William Walstad and associate editors Georg Schaur and Sam Allgood for their comments and suggestions on an earlier version of this paper. We also thank Yu Chen, Ting Dai, and Qingyun Shen for their comments on earlier versions of this paper, as well as the Undergraduate Research Project team at China Agricultural University for excellent research assistance.

Notes

¹Consistent with this argument, Sacerdote (2001) found no peer effects for economic students’ GPA at Dartmouth College.

²See http://www.cuaa.net/cur/2014/ (accessed April 6, 2014).

³Students in the liberal arts track took Chinese, mathematics, English, and a comprehensive test on history, geography, and politics combined.

⁴See Carrell, Fullerton, and West (2009) and Brunello and colleagues (2010) for similar arguments regarding different potential peer effects in different fields of study.

⁵We did not specify a cut-off for such exclusion, but other than the economics classes we included in our analysis, all economics classes in the original data set had fewer than 10 students.

⁶While this distribution may seem at odds with the case of some American universities (Main and Ost 2014), it is similar to the case of the University of Minnesota (Chen and Okediji 2014).

⁷Note that we do not control for administrative-class dummies for noneconomics students in table 2 because many noneconomics students are the only students who took economics courses in their respective majors. Given this situation, many observations will be dropped out of the regression if we controlled for administrative-class dummies.

⁸To determine if the impacts of the mean roommate CEE score found here are driven by a “leadership” effect, we also examined the impacts of the maximum roommate CEE score. When both the mean roommate CEE score and the maximum roommate CEE score are included in the regression models, the impacts of roommates’ mean CEE score remain close to those presented in table 6 (although not as significant), and the impacts of roommates’ maximum CEE score are much smaller and are not statistically significant at any conventional level. Thus, in the rest of this article, we focus on the impacts of roommates’ mean CEE score.

Additional information

Funding

This project is partially supported by Chinese Universities Scientific Fund #2014XJ013.