Common Knowledge in Coordination Games

How to cite

Chen, Yan; Riyanto, Yohanes E.; Chen, Roy. 2017. "Common Knowledge in Coordination Games." AEA RCT Registry. AEARCTR-0002406. https://doi.org/10.1257/rct.2406-1.0

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

Yan Chen* , Yohanes E. Riyanto , Roy Chen

* Primary investigator

Registry ID
AEARCTR-0002406
DOI
https://doi.org/10.1257/rct.2406-1.0
First registered
Last updated
Status
In development
Trial dates
2017-09-04 – 2018-05-31
Countries
Private
JEL codes
C71, C91, D71
Registry record
http://www.socialscienceregistry.org/trials/2406

Abstract

In the replication study of Camerer et al. (2016), Chen and Chen (2011) did not replicate. In this study, we seek to understand reasons which might have led to the non-replication. One protocol deviation in the replication was that the replication author did not read the instructions for the coordination game aloud, thus removing the common knowledge condition from the minimum effort coordination game. We conjecture that common knowledge is critical to establish efficient coordination with ingroup members. In a series of laboratory experiments, we vary the experimental condition by conducting two treatments with and two without common knowledge.

Intervention

To test the effects of common knowledge in coordination games, we employ a 2*2 between-subject factorial design. Along one dimension, we vary whether we establish common knowledge by reading the instructions for the minimum effort game aloud. Along the other dimension, we implement the original ingroup versus outgroup treatments.

Experimental design

To test the effects of common knowledge in coordination games, we employ a 2*2 between-subject factorial design. Along one dimension, we vary whether we establish common knowledge by reading the instructions for the minimum effort game aloud. Along the other dimension, we implement the original ingroup versus outgroup treatments.

Since each session consists of 12 subjects, we use 168 subjects, or 14 independent sessions to replicate the original study.

We also run an identical number of sessions following the replication protocol, i.e., without the common knowledge.

Design details

Randomization method
Randomization done in office by a computer.
Randomization unit
Each subject is randomized into one of two sessions. Sessions are randomized into one of four different treatments.
Clusters
28 sessions, 7 per treatment.
Observations
336 subjects
Arms
72 students per treatment.
Primary outcomes
effort level; communication quantity in the chat stage;

Cited by

No citation data available for this registration yet.

Keywords