Frederick Mosteller celebrated his 85th birthday late last year but remains professionally active and involved in several research projects. For a extensive review of many of his contributions, see the special volume, A Statistical Model [4], prepared in his honor. John Tukey's biography of Fred (as he is known to his friends, collaborators, colleagues, and students) in that volume is especially noteworthy.

Fred was born Charles Frederick Mosteller in Clarksburg, West Virginia, on December 24, 1916 (a birth-date he shares with my mother, and a fact perhaps worthy of examination in light of Fred's interests in the statistical analysis of coincidences) [3]. The family later moved to the Pittsburgh area where Fred attended Schenley High School, and later Carnegie Institute of Technology (now Carnegie Mellon University). In college, he was interested in mathematics and in particular, in combinatoric problems. This led him to the statistician Edwin C. Olds, who in turn led Fred into the field of statistics. Fred completed his Sc.M. degree at Carnegie Tech in 1939, and then enrolled at Princeton to work on a PhD with Samuel Wilks. Fred not only worked with Wilks and others in a wartime research group (see the discussion in [15] ), but he also assisted Wilks in his role as editor of the Annals of Mathematical Statistics. Fred was awarded the ASA Samuel S. Wilks Award in 1986. It was at Princeton that Fred began his lifelong interaction and collaboration with John Tukey, described in part in [2].

Fred met his wife Virginia when he was a college freshman (they rode the same streetcar that Fred took from Wilkinsburg into the campus each day), and they were married in 1941 just as the statistical war research effort was gearing up. Fred accepted a position in the Department of Social Relations at Harvard University in 1946, and remained on the faculty of the university in various positions for the rest of his career. Fred and Virginia moved to their home in Belmont after the birth of their son Bill in 1947. Their daughter Gale was born in 1953.

Fred became professor of mathematical statistics at Harvard in 1951 and led the effort to create the Department of Statistics in 1957. He served as the department's first chair, from 1957 to 1969. I remember my first meeting with Fred in the fall of 1964 as I was entering graduate school. Fred took me to lunch at the Harvard Faculty Club, where he insisted that I try the horse steak. As I was busy chewing, he shared with me some barely decipherable handwritten notes from Tukey on assessing probability assessors, a problem that was to be my first project as one of Fred's research assistants. Although we wrote up the results of this effort as a joint technical report, I confess it was not until a collaboration on the topic with Morrie DeGroot some 18 years later that I came to understand and appreciate both Fred's and John's insights into the problem.

As I was completing my PhD under Fred's supervision in 1968, he organized a group of us to develop a book built around the recent developments in categorical data analysis, especially linked to the use of log-linear models. This project ultimately produced "Discrete Multivariate Analysis: Theory and Practice"[1]. He was the guiding light behind the project and our constant editor and sometimes contributor, but in typical Fred fashion he insisted that only Yvonne Bishop, Paul Holland, and I be listed as "authors." Ultimately, we were able to convince him to let us acknowledge his efforts by listing him as a "collaborator" on the title page.

During his years in the statistics department, Fred formally supervised 17 PhD dissertations, but he served on the committees of countless others in statistics, social relations, and other parts of Harvard. Over the years he was always available to offer comments on works in progress and unpublished manuscripts and wise students took advantage of his generosity.

Fred later served as chair of two other departments, Biostatistics and Health Policy Management, both in the School of Public Health, and he also taught courses in the Harvard Law School and the John F. Kennedy School of Government. On retirement in 1987, Fred maintained his office in the Department of Statistics where he continued with his usual array of multi-disciplinary projects, almost as if there had been no change.

Fred's bibliography is astounding; it contains 57 books, 365 papers in books and journals, 36 miscellaneous publications, and 25 reviews. And, not surprisingly, many of these were coauthored or coedited by over 200 other individuals. My personal favorite is Fred's 1964 classic book with David Wallace, "Inference and Disputed Authorship: The Federalist," [12] which was republished in 1984 in expanded form as [13]. The intriguing analyses of the The Federalist Papers presented by Mosteller and Wallace include one of the first major uses of Bayesian methods and they provide an early exposition of Laplace's method for approximating distributions.

Among Fred's varied methodological research interests and contributions that can be found in his papers and books are: inefficient statistics, sampling, Bayesian methods, paired-comparisons, the jackknife, statistics in sports, contingency table analysis, EDA, randomized experiments, meta analysis, and research synthesis.

Depending on how one measures collaboration, John Tukey is Fred's most extensive collaborator; but others prominent on the list include Thomas Chalmers, John Gilbert, David Hoaglin, Bucknam McPeek, and Fred's long-time research assistant, Cleo Youtz. Other collaborators of note include: John Bailar, Bill Cochran, Persi Diaconis, Milton Friedman, Bill Kruskal, Pat Moynihan, Jimmie Savage, Judy Tanur, Alan Wallis, Sam Wilks, Charlie Winsor, and Gale Mosteller (his daughter). And I am pleased to be included on the list.

Beginning in the 1950s, Fred helped lead an effort to bring probability and statistics to American high schools. He was instrumental in producing teachers' manuals and this led to one of the early elementary statistics texts, "Probability with Statistical Applications" [11]. Fred used a version of this book as the text for his pioneering 1961 televised course on NBC's "Continental Classroom," which introduced him to students across the nation, young and old. Many textbooks on varied topics followed.

When the American Statistical Association set up a joint committee with the National Council of Teachers of Mathematics (NCTM) in the 1960s to change the statistical content of the secondary school mathematics curriculum, Fred led the effort once again. He helped to organize, and goaded others into contributing to, the preparation of the ASA-NCTM Committee's early products, including the 4-volume collection, "Statistics by Example" [7] [8] [9] [10] and "Statistics: A Guide to the Unknown." [14] which has now appeared in multiple forms and multiple editions.

Fred has always been an organizer and this talent was recognized by many different societies and other organizations who came to Fred for help with projects as well as to fill leadership positions. Among the societies he has led as president are (in approximate chronological order): the Psychometric Society, the American Statistical Association, the Institute of Mathematical Statistics, the American Association for the Advancement of Science, and the International Statistical Institute. Over 30 years ago, he served as vice-chair of the President's Commission on Federal Statistics, which led to the creation of the Committee on National Statistics at the National Research Council.

Recognition of his accomplishments has come in many forms. Fred had received honorary PhDs from the University of Chicago (1973), Carnegie Mellon University (1974), Yale University (1981), Wesleyan University (1983), and Harvard University (1993). He is also an honorary fellow of the Royal Statistical Society, and an elected member of the American Academy of Arts and Sciences, the American Philosophical Society, and the National Academy of Sciences.

At the age of 85, most people have been long retired and are engaged in pastoral pursuits. Yet, while Fred has been officially retired for 15 years, he remains remarkably active. So I was not totally surprised last month when perusing the table of contents of the latest issue of Statistics in Medicine, I came across a paper by Lincoln Moses, John Buehler, and guess who [5], on one of Fred's longstanding research interests, meta analysis!

Fred's 1980 The American Statistician article on presentation [6] remains "must" reading for generations of graduate students and faculty. He will remain a role model for statisticians and other scientists whom he has mentored, taught, and otherwise influenced over the years.

(Tradução livre do artigo de: Stephen E. Fienberg, Department of Statistics and Center for Automated Learning and Discovery, Carnegie Mellon University)

Tales of Statisticians - Frederick Mosteller - 1916 -