The current AER (December 2011) has an interesting article by Steffen Andersen, Seda Ertaç, Uri Gneezy, Moshe Hoffman, and John A. List entitled, "Stakes Matter in Ultimatum Games."
The authors set up high-stakes Ultimatum Games in eight villages in northeast India. The stakes varied from 20 to 20,000 rupees. At the time of the study, 20,000 rupees equaled about $410, or 200 days of labor at prevailing wages). That's a lot of cash!
One of the problems with determining if stakes matter to choice outcomes is that historically there are very few low-ball offers. To try to artificially manufacture more low-ball offers, the authors added this explicit bit of framing to the instructions for proposers:
Notice that if the responder's goal is to earn as much money as possible from the experiment, he/she should accept any offer that gives him/her positive earnings, no matter how low. This is because the alternative is to reject, in which case he/she will not receive any earnings. If the responder is expected to behave in this way and accept any positive offer, a proposer should offer the minimum possible amount to the responder in order to leave the experiment with as much money as possible. That is, if the responder that you are matched with aims to earn as much money as possible, he/ she should accept any offer that is greater than zero. Given this, making the offer that gives the lowest possible earnings to the responder will allow you to leave the experiment with as much money possible. (p. 3430)
The authors point out: "This frame informs proposers that the rational decision, if both parties aim to maximize earnings, is to offer the lowest possible amount" (Ibid., emphasis added).
Note that two things are going on here. The first is that a key insight of Ultimatum Games that in many cases people do NOT consider "gaining the most money" to be most important objective. We know this because they routinely chose to punish anonymous others at a cost to themselves. But including these instructions creates a not-so-subtle framing that the objective "should" be to earn as much money as possible.
The second point is that the authors suggest that people "ought" to make "rational" decisions, that is use logic to calculate gains and losses using a consequentialist ethical approach. But people may in fact use their moral sentiments or feelings to make such decisions. By framing both the method and the goal, the authors attempt to get an anomalous result and they succeed: Their main finding is that the demand curve for justice is negatively sloped—if sacrifice becomes more costly, there will be less sacrifice.
This study is important because it demonstrates that price does matter (an important criticism of virtue ethics that we need to consider) and second, because it shows the powerful effects of framing. This is a cautionary note to all econ teachers who claim they are only doing "science" when they frame the economic question and its method (see previous post on ethical principles for the classroom).