Experience and Expertise in Strategic Behavior

点击翻译按钮获取中文摘要

In recent years experimental methods have become an indispensable component ofeconomic research. Laboratory experiments are used to test economic theory, to discoverthe properties of alternative institutional rules, and to assist in the design of new marketinstitutions. A few examples are the design of the FCC spectrum auctions, the design of anew market institution for matching medical interns with hospitals, and the design of newmarket institutions for the deregulated allocation of electrical power.Laboratory experiments in economics and game theory rest on the premise of"transference", or"parallelism:" the assumption that the behavior of subjects in thelaboratory will be the same as the behavior of people in the actual strategic encountersthat the experiments are designed to help us understand. If laboratory subjects behavedifferently, for whatever reason, this would present a challenge for experimentalresearch: theory that failed in the laboratory might perform well in the field, or vice-versa;the properties of a trading rule in the laboratory might be different from itsproperties in the field; and trading rules designed in the laboratory, and which work wellthere, may perform poorly in the field.Recent evidence suggests that the behavior of people with expertise in certain kinds ofstrategic situations may be different than the behavior of people who are "inovices," aslaboratory subjects typically are, with little or no experience in the given strategiccontext. This evidence includes recent research by Walker and Wooders in which leadingprofessional tennis players were found to make strategic decisions that were more inaccord with the theory of strategic behavior than the decisions of laboratory subjects whohad no prior experience in the strategic situation.The research project consists of experiments designed to study whether, and how, thestrategic behavior of experts differs from the behavior of less experienced and less expertdecision makers. The experiments will be conducted primarily on the World Wide Web,and they will be designed specifically to provide for the development and identificationof varying levels of expertise and the means to identify differences in behavior that areassociated with differing levels of expertise.The results of this research promise to provide a better understanding of the role ofexpertise in strategic behavior and of the strategic contexts in which parallelism can beexpected to hold, and to help identify the settings in which Web-based experiments aresuperior and those in which laboratory experiments are likely to be superior.