Peer selection and performance - A field experiment in higher education

How to cite

Rilke, Rainer Michael; Fischer, Mira; Yurtoglu, Burcin. 2018. "Peer selection and performance - A field experiment in higher education." AEA RCT Registry. AEARCTR-0002757. https://doi.org/10.1257/rct.2757-1.0

Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.

Rainer Michael Rilke* , Mira Fischer , Burcin Yurtoglu

* Primary investigator

Registry ID
AEARCTR-0002757
DOI
https://doi.org/10.1257/rct.2757-1.0
First registered
Last updated
Status
On going
Trial dates
2017-11-08 – 2018-04-02
Countries
Private
JEL codes
D01, C93
Registry record
http://www.socialscienceregistry.org/trials/2757

Abstract

We investigate how individuals choose their peers and how endogenous peer selection influences performance. In a field experiment with students of a business school, we vary whether study groups are put together randomly, or whether students can choose their group partner endogenously. With the help of pre-elicited educational performance as well as demographic data, we are able to trace how individuals choose their partner and how these choices influence their group and individual performance during the course. Besides average performance, we analyze how the selection procedure influences the distribution of performance. We measure performance on three consecutive events. The groups' performance is measured by their score on two different exercise sheets during the semester and individual performance is measured by a student's outcome on the final exam. Students only learn about their performance after the final exam. Secondary outcomes of interest are ability beliefs, satisfaction, and social preferences.

Intervention

Students are randomly assigned to one of two exercise classes. In one class they are randomly matched with a peer. In the other class, matching is endogenous and students choose the peer themselves. In both classes, these groups of two solve the same two exercise sheets together and receive a common grade for each. Then every student individually writes the same exam. There is no intermediate feedback and students only learn about their performance a few weeks after the exam.

Experimental design

The experiment's setting is a business school in Germany. All students enrolled in a bachelor's program have to participate in the same microeconomics class during their first semester. The class is taught in three groups: two groups for students in the regular track and one group for students in the international track. Only students in the regular track are part of the experiment. Students do not know that an experiment is taking place.
Students in the regular track are randomly assigned to one of two groups. In one of those groups students are randomly matched with a peer, in the other group matches are formed endogenously. Students have to solve two exercise sheets with their peer during the semester and receive a common grade for each sheet, which each makes up 15% of the final grade. In the end of the semester, each student individually writes an exam which makes up 70% of the final grade. Immediately after the exam, students are asked to fill in a questionnaire that elicits demographic information, ability beliefs, satisfaction and social preferences. At the time of this registration the intervention has already started but the final exam has not been written and the exercise sheets have not been graded. Thus registration took place before any outcome data became available.
We received permission to run the experiment from the business school's academic director as well as its program manager. None of the institutions we are affiliated with has an institutional review board.

Design details

Randomization method
Students are randomly assigned to an exercise class by a coin flip. Within the random assignment class, students are assigned a peer with the help of a random draw by a spreadsheet program.
Randomization unit
Students are individually randomized into one of the two exercise classes. Within the exercise class with random peer assignment, students are individually matched with a peer. The type of peer assignment was thus clustered within class.
Clusters
2 classes
Observations
In total, we have 192 students (96 groups).
Arms
96 students (48 groups) in the class with endogenous peer selection, 96 students (48 groups) in the class with random peer selection
Primary outcomes
Performance in two exercise sheets (group performance) and the final exam (individual performance) in the subject of microeconomics. We analyze treatment effects both on average performance as well as on the distribution of performance.
Secondary outcomes
ability beliefs, satisfaction, social preferences

Cited by

No citation data available for this registration yet.

Keywords