Groupsize’s influence on punishment in public good game. : freeriding in social groups; are there any diff erence in smaller versus larger groups?

Abstract

This thesis is concerned with a topic in behavioral economics that is a relatively new and fast growing area of study. An area close tied up with psychology since we are looking into how people behave under certain scenarios or settings.

This thesis is built up by motivational and theoretical discussion, before looking at the data collected during two experiments using Z-Tree software by Urs Fischbacher. For regression I used Stata 9.0.

The experimental setting I used were a variant of a so-called public good game with punishment. In this game the participant contribute to a common good and get something in return from this good. It is constructed such as to give a conflict between individual and group interests. After the contribution it is possible to punish other players on the background of their and everybody else’s contribution behavior.

The motivational section contains short history of public good game and behavioral economics as well. One important aspect of my experiment is the punishment. Since I’m doing a public good game where participants contribute to a common good, I have opened up for punishment by revealing contributions of individuals. This revealing is of course anonymous, but participants can target punishment based on these contributions. This will be rigorously explained in main body of text.

My main motivation for this thesis is to look into if groupsize has an effect on how much people punish each other. Therefore, I constructed two treatments; one where participants could punish all other participants (16-1) and one were they could only punish three other assigned to same group (16-4). The contribution group was equal in both treatments, i.e. all contributed to the same common good/pool. This way I ensured that there was only one difference between the two treatments.

My hypothesis is as follows: “For any given individual contribution by individual , this individual is punished less by others in the 16-1 treatment than in the 16-4 treatment.”

Then there is a chapter of how we conducted the experiment. We held the experiment in Norwegian, but all the rules and instructions are reproduced in English. In the appendix, I have enclosed a Norwegian version of the instructions as presented to the participants. All participants got a handout of the instructions in addition to an oral presentation by reading the exact content by my assistant. At the end of this chapter, a more mathematical presentation of the game is given and a clarifying part of the difference between the two treatments as mentioned above.

A theoretical chapter explaining alternative behavioral traits comes after this. Here I look at Homo economicus, as a selfish and rational actor with preferences for only maximum economic outcome. I compare this with other preferences that can explain why people contribute and punish in this game setting.

The result chapter is somewhat limited, since one of my treatments failed. However, I did a simple linear regression to see if there was any effect of groupsize on punishment. The lack of complete dataset resulted in inconclusive outcome, but I got an indication that my hypothesis may be right. This can be a motivation for further studies into this problem; disintegrating of individual punishment in increasing groupsize. I also included some discussion of what went wrong as to enlighten those who may intend to use the same setup as I did.