Tuesday, January 18, 2011

Classroom Game Theory Experiments

Today is exam day for my AP Microeconomics students. Many of them I also had for AP US History. That means they've been in my class a year and half, five days a week, an hour and half a day. So on our last day of instruction I asked them to give me some anonymous feedback. Along with a variety of positive and negative comments, one thing mentioned was how much they enjoyed our in-class game theory experiment done earlier in the semester. Since they enjoyed it so much, I figured you might too.

The first half of class was spent discussing rationality, utility-maximization and, human selfishness. Then the students were divided into partners to complete a series of economic experiments. All experiments were done during the lunch periods that coincided with our class. There were 8 experiments total, 2 different versions of 4 basic game theory strategic situations. I gave each pair of students a sheet of paper with quotes around what they should say to each participant. Because of time and candy constraints, the sample size for each experiment is ten. Here they are the descriptions and the results:

Experiment #1 (Dictator Game):

"You are taking part in an experiment. You are one of two players. You will not know the identity of the other player nor will they know your identity. There are 20 M&M's in this bag and you are responsible for proposing how it should be divided between you and the other player. The second player has no option but to accept your proposal. The game will only be played one time. Do you understand?" "How do you propose the M&M's be split?"

Average division: 11/20 taken by the decision maker.

Interpretation of the data: A perfectly selfish and rational person would take all 20 candies. The fact that this only happened once implies many desire to be "fair".

Experiment #1.5 ((Dictator Game with Friend):

The students were given similar directions above. The two differences were it should be done with a partner of their choice (offerer chosen at random) and tell them the game will be repeated.

Average division: 8/20 taken by the decision maker.

Interpretation of the data: Again, a perfectly selfish and rational person would take all 20 candies. The decrease in the amount taken by the decider implies their desire to be fair increases when the eyes of their partner are on them. Also, by telling them the game will be repeated, they have hopes their partner will be fair in the future.

Experiment #2 (Ultimatum Game):

"You are taking part in an experiment. You are one of two players. You will not know the identity of the other player nor will they know your identity. There are 20 M&M's in this bag and you are responsible for proposing how it should be divided between you and the other player. The second player can then either accept or reject your proposal. If it is rejected neither player receives anything. If the second player accepts, the money is split according to the first proposal. The game will only be played one time. Do you understand?" "How do you propose the M&M's be split?"

Average division: 11/20 taken for the decision maker.

Interpretation of the data: I expected the amount taken by the decider to decrease out of fear their partner might reject their decision. Interesting, the incentive made very little difference. Perhaps psychologist Barry Schwartz is on to something in his critique of economics in this TED Talk.

Experiment #2.5 (Ultimatum Game with Friend and Repeated):

The students were given similar directions above. The two differences were it should be done with a partner of their choice (offerer chosen at random) and tell them the game will be repeated.

Average division: 9/20* taken for the decision maker.

Interpretation of the data: As expected, the close proximity of the partner and the threat of future games decreased the amount taken.

Experiment #3: (Earned Credit/Steal or Give Game):

"You are taking part in an experiment. You are one of two players. You will not know the identity of the other player nor will they know your identity. To begin the experiment you and the other player must first earn each of your bags of 10 M&M's (20 total) by both of you completing the maze below.

The student should now complete the maze.

You are now responsible for proposing how it should be divided between you and the other player. You can choose to keep your ten or give or steal any amount to or from the other player. The second player has no option but to accept your proposal. The game will only be played one time. Do you understand?" "How do you propose the M&M's be split?"

Average division: 16/20 taken for the decision maker.

Interpretation of the data: This resulted in the largest amount taken by the deciding partner. My explanation is that the earning of the bag of M&M's gave them the authority to take more. Their moral math was changed when they asked to do the puzzle. The fact that the candies were split into two bags, one supposedly earned by another player, didn't seem to matter.

Experiment #3.5: (Earned Credit/Steal or Give Game with Friend and Repeated):

The students were given similar directions above. The two differences were it should be done with a partner of their choice (offerer chosen at random) and tell them the game will be repeated.

Average division: 11/20 taken for the decision maker. 

Interpretation of the data: Yet again we see having the partner nearby changes the decision. Like a bird responding to a scarecrow, watching eyes make a difference.

Experiment #4 (Competitive Ultimatum Game):

"You are taking part in an experiment. You are one of 3 proposers and there is also 1 responder. You will not know the identity of the other players nor will they know your identity. There are 20 M&M's in this bag and you are responsible for proposing how it should be divided between you and the responder. If the responder accepts your offer, you wil receive the portion you suggested. If the responder rejects your offer, and takes someone elses, you will receive nothing. You are eseentially competing with the 2 other proposers to get the responder to accept. Do you understand?" "How do you propose the M&M's be split?"

Average division: 9/20 taken for the decision maker.

Interpretation of the data: When asked to bid against other decision makers, the number taken drops. Interestingly, it doesn't drop to 1. My guess is no one 1 M&M and so the risk is worth it.

Experiment #4.5 (Competitive Ultimatum Game with Enemy):

The students were given similar directions above. The two differences were it should be done with partners of their choice (offerer chosen at random) and tell them the game will be repeated.

Average division: 7/20 taken for the decision maker.

Interpretation of the data: Even though the bids were made privately on a sheet of paper, the sheer fact that their competitors were nearby made this the lowest amount taken. I believe the inherent victory (think bragging rights), led to lower amount taken.


For homework the students were asked to answer the following questions for each experiment:

1) What would you do?
2) What would an economist say you should do?
3) What would Jesus (or your preferred deity) do?

*In Experiment 2.5, there was an outlier who offered all of his M&M's. After I found out that the decision maker was a school administrator, my first explanation that they didn't want to seem selfish in front of my students. However, it turns out for health reasons they are not allowed to eat chocolate candy. So my initial question of "Do you like M&M's" (to ensure it was something they would desire) didn't work. They liked them, they just couldn't have them.

No comments:

Post a Comment

LinkWithin