Social Learning in Groups: an Experimental Study

How to cite

Agranov, Marina; Strack, Philipp; Tamuz, Omer. 2018. "Social Learning in Groups: an Experimental Study." AEA RCT Registry. AEARCTR-0003315. https://doi.org/10.1257/rct.3315-1.0

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

Marina Agranov* , Philipp Strack , Omer Tamuz

* Primary investigator

Registry ID
AEARCTR-0003315
DOI
https://doi.org/10.1257/rct.3315-1.0
First registered
Last updated
Status
In development
Trial dates
2018-10-01 – 2019-11-01
Countries
Private
JEL codes
C9, D83
Registry record
http://www.socialscienceregistry.org/trials/3315

Abstract

In this project, we use controlled laboratory experiments to investigate the aggregation of information in groups and the efficiency properties of this process. In particular, we construct a natural setting in which we will study how the size of the group, the strength of the signals, and information structure affect the degree to which information contained in the private signals is aggregated through mutual observation of others' actions.

Experimental design

In the experiment, the groups of different size will play the finite-horizon game based on the environment studied in Harel, Mossel, Strack and Tamuz (2018) paper "Groupthink and the failure of information aggregation in large groups". The treatments will differ in the number of participants in a group, the precision of participants' private signals as well as in the information available to the participants throughout the game. In addition, at the end of the experiment, we will conduct several control tasks that elicit participants' risk preferences, overconfidence and IQ measures. The additional treatments will include variants of the same game which aim to investigate the mechanism that drives outcomes observed in the baseline game.

Design details

Randomization method
Randomization is done in office by a computer
Randomization unit
Experimental sessions
Clusters
Between 200 and 400 participants
Observations
Between 200 and 400 participants
Arms
Between 20 and 100 subjects per treatment (depending on the group size)
Primary outcomes
We are interested in characteristics of the learning process and efficiency of information aggregation in different treatments. Specifically, we will investigate how fast (if at all) groups learn the true state of the world, whether or not group members converge to the same action when they observe each other's actions, whether the disagreement happens and whether its likelihood depends on the size of the group. We are also interested in documenting the groupthink phenomenon and understanding the mechanisms for its initiation and its dynamics. Moreover, we will investigate the effect of the information structures on the ability of subjects to learn the true state and impediments along this path.

Cited by

No citation data available for this registration yet.

Keywords