Laboratory Experimentation in Economics: a Methodolocical Overview*

Informal experimentation in economics goes back at least as far as Bernoulli (1738) (in connection with the Petersburg game), and formal reports of laboratory experiments as such have appeared for some time now (see e.g. Allais, 1953; Chamberlin, 1948; Flood, 1958; Friedman, 1963; May, 1954; Sauermann and Selten, 1959; Siegel and Fouraker, i960; Smith, 1962; Stone, 1958). There are even survey articles on the subject which precede this present one by more than twenty years.^ Nevertheless, it is probably only in the last ten or fifteen years that laboratory experimentation can be clearly seen to have truly begun its now steady and sustained transformation from an occasional curiosity into a regular means for investigating many kinds of economic phenomena. Incidentally, when I speak of' laboratory' experiments, I am not speaking of the location where experiments are conducted, which we will see may be in a casino or an Indian village as well as at a university. Rather I am speaking of experiments in which the economic environment is very fully under the control of the experimenter, who also has relatively unimpeded access to the experimental subjects. This distinguishes laboratory experiments from 'field' experiments, in which relatively few aspects of the environment can be controlled, and in which only limited access to most of the economic agents may be available.^ It is precisely this control of the environment, and access to the agents (sufficient to observe and measure attributes that are not controlled) that give laboratory experiments their power. The volume and variety of contemporary experimental work is now sufficiently large that, in a survey of (even) this size, I would be unable to represent it all adequately. Instead, I will aim for the more modest.