Monday, November 24, 2008

Experiments in economics

I read very quickly though Field Experiments by Harrison and List as well as Toward an Understanding of the Economics of Charity: Evidence from a Field Experiment by Landry et. al. I used to have an aversion to experimental economics but am getting around to the idea that for some interesting questions such as risk aversion, experiments may be necessary. While the survey article by Harrison and List (also in JEL 12/2004) left me confused in some parts, it convinced me that economists have learned a lot over the years.

One thing that puzzled me is that there isn't more of an intersection between survey sampling ideas (as well as randomization) with experiments. One of the things I zeroed on was the use of students in experiments (or convenience samples such as church goers). It seems that the ideas of stratification and sampling used in surveys and randomization should be introduced into experiments. (The survey article only touched on this -- they point out that even randomization has large recruiting costs but the reason these costs are borne is because randomization is in some way more acceptable than experimentation.) The biggest hurdle is cost and large scale experiementation probably hasn't made in-roads because experimental economics has not yet firmly become part of the toolkit of mainstream economists.

Perhaps with more experiments done over the Web, e.g. playing games of chance/answering questions on-line the line between surveys and experimentation will merge. I see recruiting as an ever growing problem with constant Polls, marketing surveys etc. the response rates have been declining for the past 20 years.

One last jab at JEL survey articles is that they tend to cover too much with too little depth and too much jargon. I think this article could have been made a lot simpler and narrower and hence more interesting.

No comments: