Spatial Structure of Peer Networks and Academic Achievement Based on a Random Control Trial Experiment

• Published date: 01 May 2024
• Author: Shilan Feng,Ya Tan,Zhi Wang,Qinghua Zhang
• Date: 01 May 2024
• DOI: http://doi.org/10.1111/cwe.12532

©2024 Institute of World Economics and Politics, Chinese Academy of Social Sciences

China & World Economy / 67–97, Vol. 32, No. 3, 2024 67

Spatial Structure of Peer Networks and

Academic Achievement Based on a

Random Control Trial Experiment

Shilan Feng, Ya Tan, Zhi Wang, Qinghua Zhang*

Abstract

This paper investigates whether changing the seating arrangement in a classroom can

facilitate positive spillovers from top-performing students to others, using a field experiment

conducted in a Chinese high school. Among study groups with balanced abilities, the

treatment altered the spatial distribution by assigning the two top students to seats in the

spatial center of each group. In the reference groups where students were allowed to choose

their own seating arrangements, the lowest performing were significantly less likely to sit

next to a top student than they would be under a random assignment. The results suggest

that, in the treated groups, there could be enhanced academic spillovers from the top

students. The treatment especially benefited the two lowest performing students in science

subjects. In contrast, the treatment exerted negative effects on the test scores of the two

middle-performing students, due to a disruption mechanism. The results suggest that the

spatial layout of a peer network can have a significant impact on learning outcomes.

Keywords: field experiment, peer effects, social interactions, spatial structure, test scores

JEL codes: I21, J13, R23

I. Introduction

In a well-defined peer group where peer interactions are forced or encouraged by

institutions, low-performing students could benefit from the presence of high-performing

*Shilan Feng, Assistant Professor, School of Marxism, Peking University, China. Email: carollan@pku.

edu.cn; Ya Tan (corresponding author), Assistant Professor, School of International Trade and Economics,

University of International Business and Economics, China. Email: tanya@pku.edu.cn; Zhi Wang, Associate

Professor, China Center for Economic Studies, Fudan University, and Shanghai Institute of International

Finance and Economics, China. Email: wangzhi@fudan.edu.cn; Qinghua Zhang, Professor, Guanghua School

of Management, Peking University, China. Email: zhangq@gsm.pku.edu.cn. The authors thank Jimmy Chan,

Eleonora Patacchini, Xuejing Zuo, Huailu Li, and Kevin Lang for substantive comments and suggestions.

The authors are grateful for support from the National Natural Science Foundation of China (Nos. 72203039,

72273004, and 72003036) and the National Social Science Foundation of China (No. 20CJL030). Zhang, Tan,

and Feng acknowledge the support from the Key Laboratory of Mathematical Economics and Quantitative

Finance, Ministry of Education and Peking University, China (RCT ID: AEARCTR-0007880).

Shilan Feng et al. / 67–97, Vol. 32, No. 3, 2024

©2024 Institute of World Economics and Politics, Chinese Academy of Social Sciences

68

students (Carrell et al., 2009; Li et al., 2014). Any positive spillover from high-performing

students may remain scant, however, because lowest performing students (hereafter simply

“bottom students”) tend to interact mostly with other bottom students in their peer groups

(Carrell et al., 2013). In light of the vast body of literature demonstrating the importance

of physical proximity in enhancing effective social interactions and knowledge spillovers,

this paper asks whether changing the spatial layout of a peer group in an effort to reduce

physical distance between top-performing students (hereafter simply “top students”)

and students who are not top performing can enhance their interactions and improve

educational outcomes for the latter (see Jaffe et al., 1993; Rosenthal and Strange, 2003;

Rosenthal and Strange, 2004; Arzaghi and Henderson, 2008; Bayer et al., 2008).

This paper takes advantage of several features of Chinese classrooms to investigate

whether (and how) changing the seating arrangement in a classroom can facilitate

positive spillovers from top students to others. We conducted a field experiment with

students in the tenth grade of a high school in Shanxi province in China. In Chinese high

schools, students in the same grade are divided into classes, and students in the same

class sit for most lessons in the same classroom. It is the practice of the high schools

in the province to encourage students to study in groups. In the high school that we

observed, students in the same class are divided into study groups by the teacher at the

beginning of the semester. Each group typically consists of six students who sit close

to each other in two rows of three seats during regular classes and self-study sections

over the course of the semester. The gender composition and academic performance are

kept balanced across groups (for more details, please refer to Section III). As a result,

students face similar peer environments at the group level. A student’s actual peer

environment, however, may vary depending on a group’s seating arrangement, as it may

be easier for a student to talk to someone who sits next than to someone who sits behind

or in front.

Our experiment sought to investigate whether placing top students in the spatial

center of each group can affect the academic performances of the nontop students. About

half of the groups were randomly selected as treated groups; the remaining groups were

used as reference groups. In the treated groups, the two top students (as measured by

previous test scores) were assigned to seats in the spatial center of each group block

(i.e., in the middle of each row of three). This treatment altered the spatial layout

of the group by seating a top student next to each of the nontop students, including

two bottom students and two middle-performing students (hereafter simply “middle

students”). In the reference groups, seats were arranged based on students’ preferences.

Consistent with the findings of Carrell et al. (2013), in the reference groups where

students were allowed to choose their own seating arrangements, the bottom students