St. Bonaventic Experiments: A Case Study

The experiment took place over four sessions, one at St. Bonaventure University and the remaining three at West Virginia University, from the pool of undergraduates. Subjects were recruited through an email based invitation client. In the invitation the subjects had to answer a series of questions that collected background information from the subjects (included in the appendix). On the day of the experiment subjects where invited into the lab spaced out and assigned a player id for the experiment. Subjects were then presented with a consent form and a risk evaluation. Subjects were then informed on the rules of the game and briefed on how the rules changed each round. To ensure the subjects understood the rules and how they changed they were
The objective was for the subjects to guess the color of the true envelope, which was defined by the majority card color in the envelop. If the correct envelop was selected, the subjects were awarded $10 for that round. The round would start with the subjects receiving an endowment of $5 and were given the opportunity to purchase looks from the envelop to inform their guess about the true state of the envelope. The cost of a look was $1 and if multiple looks were purchased, the additional looks were drawn with
In order to examine the effect of group size in a voting environment, after the first round group sizes changed and voting was introduced. When group sizes increased the subjects were told they were randomly placed in groups and guess of the envelope would be made based on simple majority voting, and examine group sizes of 1, 3, and 5. Based on the literature that says marginal per capita returns (MPCR) matter for voluntary contributions to the public good, we also examine a rule structure that captures group size changes while holding the MPCR constant. This was achieved with the rule structure where the group size changed but nonvoters were added to the group, again if the correct envelope was selected by the group of voters everyone would receive the benefit of a correct answer. In these rounds the groups were 1 voter and 2 nonvoters, 1 voter and 4 nonvoters, and 3 voters and 2 nonvoters. These six group sizes served as our main treatments for the experiment. As a robustness check in two of the three sessions run at West Virginia University we reran the rounds where group sizes were 1, 3, and 5, but the ratio of cards in the envelops were changed from three and two to four and one. This set of treatments was always run at the end of the session after the subjects were exposed to the main treatments of the