Participants in the three experiments conducted for this thesis were recruited from around the Boston area using two methods. Posting were made to, a community website, seeking individual interested in "participating in an exciting experiment." Additionally, recruitment posters were placed on college campuses around the Boston area.

Participant Scheduling

After signing up using the experiment scheduling system described in Appendix A, participants arrived at the Media Laboratory. Participants were asked to choose a time slot from a calendar of available sessions Monday through Friday from 10 AM to 5 PM on the hour (except for noon). If the slot they had chosen was empty they were either randomly assigned to the Interview Experiment or the Poker Experiment (experiments that required two people to show up before they could be run). In the case that someone had already taken a spot in the slot, they inherited this partner's experimental task.

Midway through the experiment (on May 13th) a system of assigning extra subjects was implemented to offset the rate at which subjects were not showing up for scheduled sessions. This system assigned a "spare" subject with each pair. I found empirically that the attendance behavior of experimental subjects can be roughly modeled as a coin flip. For any given subject there is a 50% chance of actually appearing. This meant that for pairs of subjects, there was a 1 in 4 chance of a pair actually appearing. Consequently, I implemented an over-scheduling system to place spares and increase the odds of actual pairs appearing.

The first participant to arrive was greeted and asked to wait for another subject to appear. They were told that if the another subject didn't appear by 15 minutes after the hour, then I would run them individually (which meant they would be reassigned to a third experiment -- the Quiz experiment -- which required one person to be present). If a second participant appeared, then I would escort both of them to the pair of offices used for the experiment. In the unlikely case that three subjects appeared, I would compensate the last subject to arrive $5 and ask them to reschedule for another time.

The experimenter was never blind to the condition to which participants had been assigned. A stronger future version of this experiment could involve another person who was blind to the hypothesis to perform tasks involving interacting with participants such as attaching sensors. However, the majority of the experiments were designed to be run completely by the computer, to minimize the contact with the experimenter in all but one case, which will be discussed later.


Approximately 560 participants registered for the experiments conducted. Only those subjects who arrived for their scheduled experimental session and (when paired) arrived at the same time as their partner were included in the study. In some small number of cases (less than 10) participants were not able to complete the experiment due to computer problems or interference with the experimental apparatus. These subjects were compensated, but their data was not used. This reduced the number to N=390 individuals who completed the actual experiment. Subjects who completed the experiment had to be able to make their way through all of the web pages associated with their experiment.

Figure 3.1. Histogram of study participant ages

Histogram of study participant ages

The median age reported for the dataset was 24, the mean was 27.27 with a standard deviation of 9.61. The figure above shows a histogram of the ages of the participants.

There were 199 female and 191 male participants in the dataset for all three experiments, 51% of the subjects were female and 49% of the subjects were male.

With respect to nationalities, 38 countries were represented. The largest majority (78%) were from the United States, distantly trailed by India (2%), Germany (1%), with the others representing less than 1%.

With respect to education, the majority (61%) reported an undergraduate education, followed by 31% reporting post-graduate education, and 8% reporting secondary level. No distinction was made between some undergraduate and completion of undergraduate education.