Too Good to be True - Individual and Group Decision-Making with Correlated Signals

How to cite

Fehrler, Sebastian; Hochleitner, Anna; Janas, Moritz. 2020. "Too Good to be True - Individual and Group Decision-Making with Correlated Signals." AEA RCT Registry. AEARCTR-0006323. https://doi.org/10.1257/rct.6323-1.1

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

Sebastian Fehrler* , Anna Hochleitner , Moritz Janas

* Primary investigator

Registry ID
AEARCTR-0006323
DOI
https://doi.org/10.1257/rct.6323-1.1
First registered
Last updated
Status
On going
Trial dates
2020-08-24 – 2025-03-31
Countries
Private
JEL codes
C92, D23, D71
Registry record
http://www.socialscienceregistry.org/trials/6323

Abstract

We conduct an online experiment on individual and group decision-making with correlated signals. In every
round, either one decision-maker individually or three decision makers collectively make a decision
under uncertainty that can be right or wrong. Our experiment follows a 2 (group or individual) x 2 (with or without bias) x 2 (with or without correlation) between-subjects design.
Depending on the treatment, we expect different success probabilities and communication patterns.

Intervention

We conduct an online experiment. In every round, decision-makers have to make a decision (red or blue), either individually or collectively (in groups of three), depending on the treatment. If the correct decision is taken, i.e. if the decision equals the state of the world (red or blue), each player enjoys a positive payoff. A priori, each state is equally likely and signals (red or blue) are randomly drawn from an urn representing the state of the world. Our treatments vary the bias of the decision makers, the correlation structure of the signals and whether the decisions are made individually or collectively.

Experimental design

Our experiment follows a 2 (group or individual) x 2 (with or without bias) x 2 (with or without correlation) between-subjects design. For the cell Group-No Bias-Correlation, we run an additional treatment with a message option. Hence, 9 treatments will be conducted in total.

Design details

Randomization method
Subjects are randomly assigned to treatments by the Prolific software
Randomization unit
Individuals
Clusters
1900
Observations
1900
Arms
100 for individual choice treatments 300 for group decision treatments
Primary outcomes
Decision accuracy rate
Secondary outcomes
Communication honesty

Cited by

No citation data available for this registration yet.

Keywords