Honest Communication and Cooperation in the Golden Balls Game

How to cite

Peeters, Ronald; Neururer, Daniel; Richardson, Euan. 2022. "Honest Communication and Cooperation in the Golden Balls Game." AEA RCT Registry. AEARCTR-0009587. https://doi.org/10.1257/rct.9587-1.0

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

Ronald Peeters* , Daniel Neururer , Euan Richardson

* Primary investigator

Registry ID
AEARCTR-0009587
DOI
https://doi.org/10.1257/rct.9587-1.0
First registered
Last updated
Status
In development
Trial dates
2022-06-15 – 2022-08-14
Countries
Private
JEL codes
C72, C91, D91
Registry record
http://www.socialscienceregistry.org/trials/9587

Abstract

A (somewhat counterintuitive) finding in a recently published article from Bahel et al. (2022) is that communication may lead to less cooperation in the Golden Balls Game in populations that value honesty more. In this study we test this finding empirically with the help of controlled laboratory experiments. In addition, we investigate to what extent behaviour in the Golden Balls Game is correlated with various personality traits at the individual level.

Experimental design

The experiment will be conducted online using Qualtrics and participants will be recruited using Prolific. Participants are asked to take part in 2 decision tasks, where the second task consists of two parts. In total 5 decisions need to be made in these tasks. After completing the decision tasks, there will be in total 27 short survey questions.

In task 1 (a mind game), participants are asked to imagine a number between 0 and 100. Then, we inform the subjects that the imagined number will be transferred to points (100 points equals 1 GBP) and ask the subjects to report the imagined number. After this task we divide the subjects into two groups where one group has reported relatively higher numbers than the other group (this implies that one group reported less honestly than the other).

These two groups form the basis for task 2 where we inform the subjects immediately if they are a member of the more or less honest group, which depends on the number the respective subject reported (note that we do not use the word ‘honest’ in the experiment).

In part 1 of task 2, we explain to the subjects the rules of the Golden Balls Game where two subjects are matched and have to decide simultaneously between Option X and Option Y. Depending on the options chosen by the two subjects the payoffs from this part will be determined (i.e, if both subjects choose Option X, each subject receives 100 points; if both subjects choose Option Y, each subject receives 0 points; and if both subjects choose different options, the one choosing Option Y receives 200 points, while the one choosing Option X receives 0 points).

In part 2 of task 2, the subjects have the option to communicate a (non-binding) intention regarding the option that will be chosen to the matched subjects before playing the Golden Balls Game once again. Then, the subjects make their actual option choice can contingent on the message received from the matched subject.

Finally, the subjects will be asked 27 short survey questions that elicit a separate score for the following personality traits: guilt proneness, Machiavellianism and honesty-humility.

Actual payment is according to task 1 and one of the two parts of task 2 where each part is equally likely to be selected (via computerised randomisation). These payments are processed via Prolific. Participants receive 0.01 GBP for each point and this can result in earnings up to 3 GBP based on their decisions. In addition, they receive a fixed participation fee of 2 GBP.

Before running the experiment, we will run a pilot with 40 participants. The main reason for this pilot is that we are not fully sure about the distribution of reported numbers in task 1. Therefore, in the pilot we do not divide the subjects into two groups after task 1 (we simply observe the outcome of task 1 for the sake of calibrating the main experiment). The results of this pilot could be used as a benchmark with respect to the results of the main experiment.

Design details

Randomization method
All randomisations are computerised.
Randomization unit
The randomisations are on individual level and on pairs regarding which of the two parts in task 2 is paid out.
Clusters
1
Observations
300 observations
Arms
150 participants in each of the two "honesty" groups
Primary outcomes
Test empirically a prediction from the theoretical model of Bahel et al. (2022) that states that communication may decrease cooperation in the Golden Balls Game in populations that value honesty more.

Cited by

No citation data available for this registration yet.

Keywords