Frederick  Mosteller 1916-2006

Stephen E. Fienberg, Carnegie Mellon University, and Judith M. Tanur, State University of New York, Stony Brook

Charles Frederick Mosteller (Fred to his colleagues and friends), one of the towering figures in twentieth-century statistics, was born in Clarksburg, West Virginia, on December 24, 1916, and died in Falls Church, Virginia, on July 23, 2006. Fred spent most of his childhood in the Pittsburgh area, where he attended Schenley High School and, later, Carnegie Institute of Technology (now Carnegie Mellon University). In college, he was interested in mathematics and, at an early stage, expressed interest in how to formulate problems—such as the “three dice problem”: What is the probability that the sum of the three faces equals 10?—in such a way that the answers didn’t simply involve counting. This inclination led Fred to the statistician Edwin C. Olds, who, in turn, steered him into the field of probability and statistics by teaching him about generating functions. “It was the most marvelous thing I had ever seen in mathematics. It used mathematics that, up to that time, in my heart of hearts, I had thought was something that mathematicians just did to create homework problems for innocent students in high school and college. …I was stunned when I saw how Olds used this mathematics that I hadn’t believed in…in such an unusually outrageous way.” From that point, Fred was hooked on probability and statistics. He completed his ScM degree at Carnegie Tech in 1939, and then enrolled at Princeton University to work on a PhD with Samuel Wilks.

Fred spent his entire academic career (except for leaves and sabbaticals) at Harvard University. His initial appointment in 1946 was in the Department of Social Relations, and he was the founding chair of Harvard’s statistics department in 1957. Later, he chaired two departments in the School of Public Health: Biostatistics and Health Policy and Management.

Fred’s bibliography lists 65 books, nearly 350 papers in books and journals, 41 miscellaneous publications, and 26 reviews. Many of these were in collaboration with others, a style of work he was enormously comfortable with. He encouraged collaboration, setting up groups to work on various projects and, by shouldering more than his share of the responsibility, inspiring others to turn out more and better work than they might have otherwise. It is impossible to include a full intellectual biography here; more details appear in the introduction to Selected Papers of Frederick Mosteller, edited by Stephen Fienberg and D. C. Hoaglin, and the Amstat News article “Frederick Mosteller: a Brief Biography,” by Fienberg. His forthcoming autobiography, My Statistical Life, gives many more details, including his early work on the 1948 pre-election polls, the Kinsey Report, psychological learning models, the disputed Federalist Papers, and the Coleman Report on equal education opportunity. Many of Fred’s accomplishments are chronicled in A Statistical Model: Frederick Mosteller’s Contributions to Statistics, Science, and Public Policy, a volume assembled by colleagues in honor of his 70th birthday.

Here, we concentrate on Fred’s activities directly connected with the American Statistical Association. Of course, he was an ASA Fellow; he received the Wilks award in 1986; he was president of the ASA in 1967; he received the Founders Award in 1992; and he led the very successful Joint ASA/National Conference of Teachers of Mathematics Committee on the Curriculum in Statistics and Probability. As chair of the ASA/NCTM Committee, Fred led a two-pronged attack on the problem of getting statistics and probability included in the high school curriculum. He saw the first task as one of persuading those responsible for setting the curriculum—members of school boards and their constituents—of the value of statistics and probability and the contributions of these disciplines to the advancement of the biological, political, social, and physical sciences, as well as their usefulness in everyday life. To make this case, the committee decided to produce a book of essays recounting statistical success stories in a style readable by laypeople. Although faced with skepticism that serious writing at that level was possible, Fred and the other committee members wrote examples to show it could be done, solicited essays from statisticians working on problems of interest to the general public, edited and often rewrote the essays, and, in 1972, published Statistics: a Guide to the Unknown. The book was very well received, and Fred was the leader in publishing a slightly revised second edition in 1978 and an almost completely new third edition in 1989. A completely new fourth edition, produced by a new team of editors, recently appeared.

As the second prong of its attack, the committee provided real and interesting instructional material for teachers in high school to use in courses in statistics and probability. To this end, the committee prepared four volumes under the general title Statistics by Example, with the intention that teachers should feel free to use these materials in the classroom—books designed for “plagiarism” in a good cause. (This is a rather commonplace idea in our current age of web searches and open-source programs; it was almost unheard of in the 1970s.) The subtitles of the volumes give some idea of Fred’s mapping of the field of statistics: Exploring Data, Detecting Patterns, Finding Models, Weighing Chances.

Several accomplishments of Fred’s tenure as ASA president are particularly noteworthy. It was during his watch that ASA made the decision to employ a full-time executive director. This change from the previous arrangement of having a statistician look after the office as a part-time job had enormous implications for the increased range of services the ASA could offer to its members and to the profession. It was the first step toward the broad functions the ASA offers today. The step also had (and continues to have) financial implications: It resulted in a dues increase (to $18 a year). Also during Fred’s watch, the Journal of the American Statistical Association (JASA) split into two sections, one on Theory and Methods and the other on Applications. This division was inspired by the perceived low rate of submission of papers on applications, of expository papers, and of review articles. The goal of broadening the coverage of JASA is a continuing one for JASA editors and for the ASA. Of the sections going into effect in 1967, Fred wrote in recognition of the unity of the field of statistics and his own conviction that theory and applications feed off each other, “Classification will be based on the primary contribution of the paper. …Precise delineation is not important and borderline cases may appear in either part.”

Fred’s 1967 presidential address, published in JASA, was groundbreaking, representing the first time such an address dealt with serious issues in statistical methodology. Fred spoke about the analysis of contingency table problems, including those studied by his students and colleagues, many of whom provided draft materials for possible inclusion. Several of the components derived from work associated with the National Halothane Study, a massive National Research Council project studying the safety of anesthetics and whose statistical component Fred coordinated. This was one of the first expositions on log-linear-model methods, as they were to emerge and ultimately be described in Discrete Multivariate Analysis: Theory and Practice, a volume Fred influenced and edited repeatedly, but of which he chose not to be a coauthor so as to let others gain professional recognition.

From the 1960s onward, Fred was engaged in a multiplicity of group projects, sometimes involving a handful of colleagues and students and other times involving large faculty groups meeting as a seminar. Many of these activities took the form of research evaluation and synthesis, especially in the areas of medicine and public health (e.g., with respect to different kinds of surgery). Products from these efforts included many papers and books that can be found in Selected Papers of Frederick Mosteller.

Fred was always a fine teacher and very involved with his teaching (a detailed discussion of his teaching activities appears in A Statistical Model: Frederick Mosteller’s Contributions to Statistics, Science, and Public Policy). A much-cited article on classroom and platform performance, “Classroom and Platform Performance,” appeared in The American Statistician in 1980 and was reprinted in Amstat News in 2002. Though the technology referred to (e.g., be sure you have blackboard chalk with you) is somewhat outdated, the spirit of the advice (e.g., be prepared; be sure you’re keeping the students awake—by opening windows, if necessary) remains useful and helpful for modern students, teachers, and researchers.

Recognition of his accomplishments came in many forms. Fred received honorary degrees from the University of Chicago (1973), Carnegie Mellon (1974), Yale (1981), Wesleyan (1983), and Harvard (1993). In addition to his ASA fellowship, he was a fellow of the Institute of Mathematical Statistics; an honorary fellow of the Royal Statistical Society; an honorary member of the International Statistical Institute; and an elected member/fellow of the American Academy of Arts and Sciences, the American Philosophical Society, the Institute of Medicine, and the National Academy of Sciences. He also served as the Committee of Presidents of Statistical Societies’ R. A. Fisher Lecturer on methods for studying coincidences, discussing work done in collaboration with Persi Diaconis.

During the 1990s, despite his putative emeritus status, Fred continued his interests in applications and methodology of statistics. Besides his work on research synthesis, he remained an advocate for evidence-based decisionmaking for policymaking. He was particularly enthusiastic about the Tennessee Class Size experiment, which actually randomized children into small vs. large classes in the early grades and traced the effects of that differential treatment through their years in public school, finding that the beneficial effects of being in a small class persisted for many years, even after students were integrated into normal-sized classes.

Honors continued to pour in, even after Fred’s move to Virginia in 2004. The Campbell Collaboration honored him for his contributions to meta-analysis and systematic reviewing and inaugurated the Frederick Mosteller Award for distinguished contributions to research synthesis. In 2005, he received the first Peter H. Rossi Award for Contributions to the Theory or Practice of Program Evaluation from the Association for Public Policy Analysis and Management.

Additional Reading
Albers, D.J. and Reid, C. (1990). “[An Interview with] Frederick Mosteller.” In More Mathematical People: Contemporary Conversations. Harcourt Brace Jovanovich, 241–260.

Bishop, Y.M.M.; Fienberg, S.E.; and Holland, P.W. (with contributions by R.J. Light and F. Mosteller). (1975). Discrete Multivariate Analysis: Theory and Practice. MIT Press: Cambridge, MA.

Bunker, J.P.; Forrest, Jr., W.H.; Mosteller, F.; and Vandam, L.D. (1969). The National Halothane Study, Report of the Subcommittee on the National Halothane Study of the Committee on Anesthesia, Division of Medical Sciences, National Academy of Sciences—National Research Council. U.S. Government Printing Office: Washington, DC.

Diaconis. P. and Mosteller, F. (1989). “Methods for Studying Coincidences.” Journal of the American Statistical Association, 84: 853–61.

Fienberg, S.E. (2002). “Frederick Mosteller: a Brief Biography.” Amstat News, 303: 10–12.

Fienberg, S.E. and Hoaglin, D.C. (eds.). (2006). Selected Papers of Frederick Mosteller. Springer-Verlag: New York.

Fienberg, S.E; Hoaglin, D.C.; Kruskal, W.H.; and Tanur, J.M. (eds.). (1990). A Statistical Model: Frederick Mosteller’s Contributions to Statistics, Science, and Public Policy. Springer-Verlag: New York.

Mosteller, F. (1967). “The President Reports: Three Major ASA Actions.” The American Statistician, 21: 2–4.

Mosteller, F. (1968). “Association and Estimation in Contingency Tables.” Journal of the American Statistical Association, 63: 1-28.

Mosteller, F. (1980). “Classroom and Platform Performance.” The American Statistician, 34: 11-17.
Mosteller, F. (1995). “The Tennessee Study of Class Size in the Early School Grades.” The Future of Children, 5: 113–127.

Mosteller, F. (2007). My Statistical Life. Springer-Verlag: New York.

Mosteller, F.; Kruskal, W.H.; Link, R.F.; Pieters, R.S.; and Rising, G.R. (eds.) (1973). Statistics by Example (Volume 1: Exploring Data, Volume 2: Weighing Chances, Volume 3: Detecting Patterns, Volume 4: Finding Models) Addison-Wesley: Reading, MA.

Peck, R.; Casella, G.; Cobb, G.; Hoerl, R.; Nolan, D.; Starbuck, R.; and Stern, H. (eds.) (2005). Statistics: a Guide to the Unknown (4th ed.). Brooks-Cole: Belmont, CA.

Tanur, J.M.; Mosteller, F.; Kruskal, W.H.; Link, R.F.; Pieters, R.S.; and Rising, G.R. (eds.) (1972). Statistics: a Guide to the Unknown. Holden-Day: San Francisco.