His scholarly contributions to the social sciences were fundamental and wide-ranging. In addition to his work in voting theory, he made fundamental contributions to game theory, social choice theory, experimental political science, and computational economics. Besides contributing research articles to the leading academic journals, he served on the editorial boards of many journals over the course of his illustrious career.
In one celebrated paper, McKelvey showed that decisions made under the one-person/one-vote, majority-rule system of legitimate democracies do not necessarily cluster around "middle-ground" policy outcomes, as had always been assumed. Rather, such decisions are very sensitive to the details of the process, such as agenda control. As a consequence, nearly any outcome, even very unpopular ones, can result from manipulations.
For this and other major contributions to political science, McKelvey was elected to the National Academy of Sciences in 1993. He accumulated numerous other honors, including election to the American Academy of Arts and Sciences in 1992, and fellowship in the Econometric Society, and was named Sherman Fairchild Distinguished Scholar by Caltech in 1978, the year before he joined the Caltech faculty.
McKelvey was a pioneer in the use of laboratory experiments to test theories of voting and other group behavior, and the application of computational techniques to understanding strategic behavior. Some of his experimental work, done in collaboration with Caltech colleagues, investigated the effect of different voting rules on the accuracy of jury verdicts, the effect of polls on election outcomes, and impasses in negotiations and bargaining.
In the last few years, he was developing a general statistical theory of games, called quantal response equilibrium, and designing experiments to test its predictions about behavior. He has published widely on this topic in the leading academic journals. His work on computation in game theory began with Gambit, his celebrated computer program, which finds numerical approximations of solutions to games.
Most recently, he initiated a contest based on a Turing tournament, designed to improve our ability to predict how people will behave in strategic situations. The tournament, to be conducted this summer, invites leading scholars in the fields of economics and game theory to compete for a large prize, which will be awarded to the theory that best matches laboratory human behavior.
Professor McKelvey was not only an innovative and creative scholar, but will be remembered by his students and colleagues as a fine educator and devoted teacher. He was highly sought after as a PhD advisor, and spent countless hours working with his students. Many of these former students now hold professorships at the leading universities of the world and carry on his approach to social scientific inquiry based on formal mathematical logic combined with rigorous empirical and laboratory testing.
Born April 27, 1944, McKelvey graduated from Oberlin College with a bachelor's degree in mathematics in 1966. He earned a master's degree in mathematics from Washington University in St. Louis in 1967, and a master's and doctorate in political science from the University of Rochester in 1970 and 1971, respectively. After serving on the faculties of the University of Rochester and Carnegie-Mellon University, he joined the Caltech faculty in 1979 as a full professor, and was awarded the Wasserman Chair in 1996.
He is survived by his wife, Stephenie Frederick, and three children, Kirk, Christopher, and Holly.
Contact: Robert Tindol (626) 395-3631