Edward J. Dudewicz, born on 24 April 1942, died on 22 September 2010 after a 6 year bout with cancer. He was Professor Emeritus of Mathematics at Syracuse University, New York. He received a bachelors degree from the Massachusetts Institute of Technology in 1963, and a masters and Ph.D. degrees from Cornell University in 1966 and 1969, respectively. His dissertation, “Estimation of Ordered Parameters,” was written under the supervision of Robert E. Bechhofer.
Professor Dudewicz was one of a handful of scientists world-wide to be elected Fellows of all of the Institute of Mathematical Statistics, American Statistical Association, American Society for Quality, and AAAS. He was Founding Editor of the Basic References in Quality Control: Statistical Techniques of the American Society for Quality, for which he received the ASQ Testimonial Award for “Leadership and Distinguished Service” from the Directors of the Society. He also was the Founding Editor of the American Journal of Mathematical and Management Sciences, currently in its 30th volume. He had over 170 publications, including 11 books, one each translated into Arabic and into Indonesian.
Dr. Dudewicz was active as instructor, researcher, editor, and consultant for over four decades. He taught at Syracuse University, Ohio State University, University of Rochester, University of Leuven, Belgium, and National University of Comahue, Argentina. His Editorial posts included Technometrics (Management Committee), Journal of Quality Technology (Editorial Review Board), Statistical Theory and Method Abstracts (Editor, U.S.A.), Statistics & Decisions (Editor), and American Journal of Mathematical and Management Sciences (Founding Editor and Editor-in-Chief).
Professor Dudewicz was internationally recognized for his solution of the Heteroscedastic Selection Problem, his work on Fitting Statistical Distributions, his development of the Multivariate Heteroscedastic Method, and his approach to the solution of the Behrens-Fisher Problem. In 1975, Professor Dudewicz with co-author S.R. Dalal published a solution of “The Heteroscedastic Selection Problem” given by Professor Bechhofer in 1954. This paper was a comprehensive 51-page analysis of the problem and its solution, including the tables needed to apply the solution. There are now many published statistical procedures which follow the method established in this path-breaking paper, and they are often referred to as “Dudewicz-Dalal type” procedures. This approach considered the general ranking and selection goal as well, and gave details for the cases of selecting the t best populations and of subset selection. Extending this idea for many multipopulation problems such as testing of hypotheses, multiple comparisons, estimation of ordered parameters, partitioning of a set of populations, confidence intervals, ANOVA and MANOVA and regression, Professor Dudewicz gave a general solution with the multivariate analogs of these problems. He and his graduate student, T.A. Bishop, in 1979 gave “The heteroscedastic method,” in Optimizing Methods in Statistics.
An important statistical problem faced by researchers in virtually any field that uses statistics is that of fitting a distribution to a set of data. Some milestones in this problem have been the Pearson system (1895), the Johnson system (1949), and the Ramberg and Schmeiser Generalized Lambda Distribution (GLD) system introduced (1972). In the years since, Professor Dudewicz and co-authors, principally Z.A. Karian, developed the GLD system further to satisfy such needs as being able to fit a distribution to any set of moments as well as to bivariate data. A comprehensive book on this subject, Handbook of Fitting Statistical Distributions with R, co-authored with Z.A. Karian appeared in print a few days after his death.
The problem of testing equality of two means when variances are not known, the Behrens-Fisher Problem, has been called the most important problem of applied statistics by Henry Scheff´e. Following two exact solutions, in 1950 by Chapman and in 1974 by Prokof’yev and Shishkin, in 1998 Professor Dudewicz and S. U. Ahmed gave a third exact solution in the paper “New exact and asymptotically optimal solution to the Behrens-Fisher problem, with tables” American Journal of Mathematical and Management Sciences. Their solution was stated to be asymptotically optimal, hence preferable to the other two solutions.
Professor Dudewicz was a generous person, one who went out of his way to help and encourage younger colleagues. Through his myriad professional activities he collaborated with many individuals who came to value him as a colleague and as a friend. His many associations with foreign colleagues were based on a genuine interest to stimulate their research. This led to many long-term collaborations and close friendships. He had a keen curiosity about their cultures and helped them with much advice. Professor Dudewicz is survived by his wife of 47 years, Patricia, three children Douglas Robert and Carolyn and five grandchildren. He will be missed by his family, friends and the statistical community. We remember him as a person full of energy, inspiration and integrity.
Zaven A. Karian
E.C. van der Meulen