Even though I was in St. Louis for Robotics this past weekend, I still made sure to work on my independent. For one, summer is drawing near, and so I need to conduct my experiment when I still have this vast pool of subjects as my disposal. However, I have also gotten to the exciting part of the project where all of my research will begin to pay off. (Though I must continue to keep in mind that my result may show nothing, which would be sad, but I cannot force correlations.) So over the week, I spent a good part of my evening practicing on my fellow robotics members. Now, of course, none of their data will be used as I am still making the final tweaks to the procedure.
Despite all of my research, there were some problems that I did not anticipate. One of the largest problems I found was the repetition of counter offers. It was a common trend for people to decide on a number to use every single time they rejected the other player’s offer. This is problematic since with no variance it becomes next to impossible to tell how implicit bias is affecting their decisions. Now, this could have been because of a couple of participants knowing about my study of game theory and therefore were purposefully more analytical. So before I move onto the final stage I also want to test this on some people I don’t know as well. However, I have a few other solutions that may alleviate the issue. One idea is to disallow repetition so that they need to use a new number each time. However, this could be problematic as it could cause people to just increment their offer. The other more complicated solution is to cue participants up to the idea of responding with different numbers, reminding them that different people may accept different offers based on their notions.
Through my practice, I also learned a variety of things that worked. First off, the instructions seem to work well, though it does often take a second or two for each subject to grok what exactly was going on in the experiment. I was able to clear up any confusion though by asking for any questions the participants might have. The time limit also worked well, moving the experiment along and not giving the participants to much time to think through their answers.
Also, for those curious, there is an actual Nash Equilibrium for the game I am using. However, given that people are not going into this with the assumption that people act strictly rational, the Nash Equilibrium has little bearing to the actual experiment. For those who have been following along all year, take a guess in the comments. I am curious if anyone can figure it out. (For a reminder about the specifics of the game: https://independentseminarblog.com/2017/04/11/procedure-update/#more-15906 [it was the second game described]).
Thanks for reading.
For those who want to read some of the studies I have been using: