One, Two, (Three), Infinity, ... : Newspaper and Lab Beauty-Contest Experiments

摘要

Experimenting with the "Beauty-contest" game through the platform offered by several newspapers allows us to explore three issues. The first is the assumption of "parallelism" between lab and field, so basic to any experimental methodology. Experimental results are influenced by what Jacob Marshak (1968) called the different costs of thinking, calculating, deciding, and acting. Large-scale experiments of the sort that can be run through a newspaper can test whether the results of lab experiments are robust to variations in sample sizes, rewards, and the different costs mentioned by Marshak. In a newspaper experiment, one is likely to encounter a population more heterogeneous than undergraduate subjects. There may be subjects with widely different costs of thought and calculation (due to different education, training, or information), different decision costs (at leisure vs. time constrained), and different costs of taking action (ready access to e-mail and fax or not). This is a richer world with less experimental control. The fact that three experiments involving thousands of subjects, run in different countries, for different newspapers, catering to different populations, yield very similar results is a clear indication that we are observing a pattern of behavior that must be quite common. In addition, this pattern is replicated in lab experiments with subject pools of undergraduate, graduate students, and economists. This indicates that the "parallelism" assumption between lab and field has been upheld.