I have been thinking about Bryan Caplan’s assertion that voters are altruistic (which I have discussed before). Caplan’s argument relies on the fact that the probability of any vote being decisive is very low, so the individual costs of voting altruistically are low, while the psychological benefits may be large. I have realized that this theory is not only empirically testable, as Bryan discusses in his book, but also experimentally testable.  Here is one experiment which would test this theory:

Gather a group of 20 people in a large room. Have them sit down and play icebreakers and otherwise get to know each other for 30 minutes and then vote by secret ballot (secret from the other people, not the experimenters) between two measures. The people choose between a measure that gives the group \$100, evenly distributed, and a measure that gives the group \$200, somewhat unevenly distributed, with some people receiving less than the \$5 they would have received from the first measure. The voting instructions explain what the reader would receive from each measure and the general distribution from each measure. After voting is completed, the sums are distributed privately. The group then socializes for another 10 minutes (they should be prohibited from discussing the voting), and then the experiment is over.

People who would receive more from the egalitarian measure than the uneven distribution measure who vote for the uneven distribution measure voted altruistically. Likewise, people who would receive more from the uneven distribution measure than from the egalitarian measure who vote for the egalitarian measure also voted altruistically. Many similar experiments could be conducted to determine more precisely the circumstances under which people vote altruistically, if at all.

I would love to run this experiment, but I don’t have the resources. I wonder if anyone has already done similar experiments.