Strangers in a not-so-strange land
For those of you who participated in this experiment, here are the results.
To recap, each of you read a description of a hypothetical interaction in which you and other person were each given 10 dollars. You (the other person) could transfer anywhere between 0 and 10 of those dollars to the other person (to you). Every dollar transferred got doubled.
This type of interaction is sometimes called a Prisoner’s Dilemma. The dilemma exists because each player does best by not transferring anything, irrespective of what the other person does. For example, if the other person sends you nothing, you maximize your earnings by keeping everything too, which, in this case would be $10. If, instead, the other person had transferred the full $10, you would still do best by keeping your $10 so you would take home a total of $30. So, each player does best by keeping everything.
Notice, though, if both people play this way, they each go home with $10. That is, they keep everything and transfer nothing. If they had both transferred all $10, they would each go home with $20.
This is a dilemma because each player does best by being selfish, but both could do much better (double in this case) if they gave away everything.
In my particular experiment, there were two variables. Variables are nothing more than what they sound like—things that vary. In this case, they vary between participants. The two variables were (1.) Question Order and (2.) Avatar Color.
In the experiment each of you was asked two questions:
- How much would you transfer to the other person?
- How much do you think the other person would transfer to you?
Roughly half of you saw the questions in the order presented above (Transfer First) and roughly half saw the questions in the other order (Receive First).
(I say roughly half instead of exactly half, because participants were assigned to different conditions based on a randomization procedure. It’s like a coin was flipped to determine whether you ended up in the Transfer First or Receive First condition. Because this process is random, there is no guarantee it will be 50-50. However, as the number of participants increases, distributions will get close to that 50-50 mark.)
While reading about the interaction and answering the above questions, there were pictures next to the text, meant to depict the interaction. These pictures included arrows for direction of a (hypothetical monetary) transfer and avatars meant to depict you and the other person. The avatars were either colored blue or red. Roughly half of you were in the Same Color condition, in which both avatars were the same color (both were blue or both were red). Roughly half of you were in the Different Color condition, in which one avatar was blue and the other red.
Wednesday, September 12, 2007
Okay, that was the basic set up. Now, when it comes to experiments in general, researchers usually have specific predictions they want to test. These predictions are derived from existing theory or gleaned from other experimental results.
In this case, there is a large body of research in social psychology on so-called in-group favoritism in Minimal Group Experiments. The basic setup is to bring subjects into the lab and assign half of them to one group and half to the other. The assignment process can be arbitrary. A rather odd, but popular one used is to ask subjects whether they prefer the paintings of Paul Klee or Wassily Kandinsky. Now, subjects play a game, like the Prisoner’s Dilemma described above, either with someone who shares their preference in painters or not. Surprisingly, subjects tend to be more generous when paired up with someone who has a similar preference.
I was interested to know whether this would extend to a web setup where there was no real group assignment. Instead, there would be a hint of similarity or dissimilarity based on the colors of the two avatars.
So, I had two specific predictions in mind when conducting this experiment.
(1.) People will offer more and expect more (hypothetical money) in the Same Color condition compared to the Different Color condition.
(2.) People will offer more when in the Transfer First condition compared to the Receive condition.
And now for the results. (Imagine a drum roll. Please!)
First off, it seems you all have a rather pessimistic attitude of your fellow man or woman. On average, you indicated you would transfer $5.80 to the other person and expected to receive only $4.33 back from the other person. We’re all saints and they’re all sinners!
I don’t really know what to make of an apparent sex difference. Men said they would offer $6.30 and expected to receive $5.00 back. Women said they would offer $5.64 and expected to get back $4.13.
Unfortunately for me, the two, main predictions didn’t hold up. There were no differences in stated transfers or expected receipts when comparing the different color conditions or the different question orders.
So, what does this all mean? I’m not sure, to be honest. With this kind of research, it could mean that the variables under investigation really don’t have any affect. It could also be that the study was not well designed and so nothing showed up. I’ll have to do some more thinking before proceeding. A good life lesson in general, I suppose.
While this kind of research may interest a few of you, many might be thinking: Is this what researchers really do with their time? Rest assured, while my PhD may turn out to be a big waste, this particular study probably is not. This study was part of a larger study designed to better understand how we think about groups. I was hoping that this Internet study would help in designing an experiment in which participants play similar games to the ones you did, but for real stakes.
Same Color
Different Color