Mosteller, Frederick, 1916-2006

Frederick Mosteller taught mathematical statistics in the Department of Social Relations and the Department of Statistics, 1946 to 1987, and was Irving Professor of Mathematical Statistics in the School of Public Health.

Frederick Mosteller, 1916- , BSc, 1938, MSc, 1939, Carnegie Institute of Technology; AM, 1941, PhD, 1946, Princeton University; Roger Irving Lee Professor of Mathematical Statistics, Emeritus, came to the Harvard in 1946; he became chair of the Department of Biostatistics at the Harvard School of Public Health in 1977 and held appointments at Harvard Medical School and the Kennedy School of Government. Mosteller later served as Chairman of the Department of Health Policy and Management from 1981 to 1987. Mosteller applied his statistical research to politics, literature, health, education, and legislation.

C. Frederick Mosteller, a native of West Virginia, received his B.S. (1938), an M.S. (1939) from the Carnegie Institute of Technology, and and also received an A.M. (1942) and Ph. D (1946) in Mathematics from Princeton University. Mosteller's entire academic career was spent as a Professor of Statistics with Harvard University (1946-1987), and he was designated a Professor Emeritus of Mathematical Statistics in 1988. He was a member of numerous academic societies, including the National Academy of Sciences, the American Philosophical Society, and the American Academy of Arts and Sciences. Mosteller's research interest was theoretical statistics and its application to education, social science, health and medicine, public policy, industry, data analysis and sports.

Dr. Mosteller married Virginia Gilroy in 1941, and they had two children, William and Gale.

A prolific and decorated scholar, a talented and hard-working administrator, and an encouraging and supportive mentor, Frederick Mosteller was one of the leading statisticians of the twentieth century. He wrote and edited over 50 books and published nearly 350 papers. He was instrumental in establishing the Statistics Department at Harvard University, where he taught and administered for over 50 years. Mosteller contributed to numerous studies in wide-ranging fields from social sciences to healthcare to education. His work was widely recognized and hailed as influential, both in its breadth and content. He was celebrated as both a scholar and leader in statistics. The Mosteller Papers are an important source for scholars in multiple disciplines, both for those interested in particular ground-breaking studies, and for those interested in the history of statistics and statistical methodology.

Charles Frederick Mosteller was born on December 24, 1916 in Clarksburg, West Virginia. His father, William Roy Mosteller, was a road builder and his mother, Helen Kelley Mosteller, was an accounting clerk and homemaker. Frederick Mosteller grew up in the Pittsburgh area and graduated from Schenley High School, where he helped form the chess team. He pursued his collegiate studies at nearby Carnegie Institute of Technology (now Carnegie Mellon University), initially focusing on engineering. Due in part to the influence of Edwin G. Olds, Mosteller changed the focus of his studies. Olds challenged the young scholar and opened his eyes to the practical application of statistics. Up to that point, Mosteller believed mathematics to be primarily a theoretical field. Olds, through his own interests in probability and statistics, demonstrated to Mosteller the application of statistics to practical matters.

During his sophomore year at Carnegie Institute, Mosteller met another individual who would influence the course of his life, his future wife, Virginia Gilroy, a student at the Margaret Morrison School for Women. The two became good friends, dated for a period of time, and married in 1941. The Mostellers later had two children together, William (born 1947) and Gale Robin (born 1953).

Mosteller received Bachelor of Science and Master of Science degrees in mathematics from Carnegie Institute in 1938-1939. He pursued further education in statistics at Princeton University, where he studied under Samuel Wilks and John Tukey. Mosteller earned a Master of Arts (1941) and Ph.D. (1946) from the Ivy League school.

During World War II, Mosteller worked for the Research Branch of the War Department and the Statistical Research Group, assisting with statistical reports in such areas as bomb delivery systems.

Mosteller first joined Harvard in 1946, beginning a 50+ year association with the prestigious university. He was hired to lecture in the Department of Social Relations. The department consisted of a highly active, intellectually curious, interdisciplinary group of social scientists from different fields -- in particular, anthropology, psychology, and sociology. It was led by Talcott Parsons, a sociologist who appreciated the use of quantitative theory and methodology.

During this early period of his professional career, Mosteller served as chief of the technical staff of a Social Science Research Council committee investigating the erroneous predictions from pre-election poling during the 1948 presidential election. The committee found that pollsters had overestimated their predictive ability and neglected the difficulties in predicting a close election. It also found the faulty nature of quota sampling and pre-elective polling. He also undertook a long, eight-year collaboration with nuclear physicist Robert Bush on developing mathematical models for learning. The culminating work from this endeavor was Stochastic Models for Learning (1955). The Bush-Mosteller work on learning theory proved instrumental in laying the foundations for mathematical psychology.

During this time, Mosteller formed a professional relationship that was to benefit him greatly. In 1952, he began working with Cleo Youtz, a research mathematician. Youtz worked closely with Mosteller for half a century until his full retirement from Harvard in 2003, assisting and contributing with his research and writing, and helping to serve at times as secretary or administrator.

While he found the atmosphere at the interdisciplinary Department of Social Relations to be stimulating, Mosteller began to call for an academic home for statisticians at Harvard. Prior to 1946, statisticians at the university were scattered across numerous departments. Statistics was treated not as a distinct discipline, but as a methodology within other disciplines. Along with statistician colleagues in the mathematics and engineering departments, Mosteller approached Provost Paul Buck in 1949 and proposed the creation of a new department of statistics. Initially told that the field was too narrow for its own department, Mosteller did not despair. After spending a year of sabbatical (1954-1955) at the University of Chicago, which did possess a statistics department, Mosteller renewed his call for a new department at Harvard. Fortunately, a new provost with a mathematics background, McGeorge Bundy, arrived at Harvard, and after creating a Committee on Applied Mathematics and Statistics (Mosteller chaired the subcommittee on statistics), the prospect of creating a new department seemed to be more likely. Mosteller received a setback when Bundy informed him that the funds were lacking to establish a new department. Disheartened, Mosteller considered a job offer from Chicago, only to be told by Bundy that a new department could be established, provided the Faculty approved. On February 12, 1957, Mosteller presented his case to the Faculty of Arts and Sciences. After a brief discussion and debate (14 minutes), the motion was approved, and Mosteller succeeded in creating a new academic department at Harvard University. Later in life, when asked how he established a new department at on old university, Mosteller replied, “Very slowly.”

As a thesis advisor, Mosteller primarily advised students who were interested in methods of data analysis. Mosteller highly valued education, enjoyed teaching, and jumped at the chance to teach mathematics and statistics to a wider public audience in 1961. He filmed a series of introductory lectures on statistics for NBC, as part of the Continental Classroom. The programs ran nationally and were used by secondary schools, colleges, and universities. Mosteller ended each program with a poignant Latin phrase: Disce ut semper victurus, vive ut cras moriturus (“Learn as if you would live forever, live as if you would die tomorrow”).

With David Wallace, Mosteller wrote a revolutionary work, Inference and Disputed Authorship: The Federalist (1964), addressing the authorship of the fifteen disputed Federalist Papers . Using Bayesian inference on texts where the authorship was known, they noted the different author’s tendency to use certain non-contextual words--conjunctions, prepositions, and articles. By noting the frequency at which certain words were used and comparing the results of the known texts with the unknown texts, Mosteller and Wallace determined that James Madison was most likely the author of the disputed texts. Their work was impressive for its scope--it was an early attempt at large-scale Bayesian applications--and it involved the use of text data-mining, a technique that became popular in subsequent statistical studies.

With an interest in social sciences and an inner drive which moved him to seek to make a difference in the world, Frederick Mosteller participated in various public policy studies throughout his career. He took part in the National Halothane Study (1966-1968), a large-scale study undertaken to determine if the anesthetic halothane was safe for public use. Collaborating with physicians and working on public health issues became an important part of Mosteller’s professional life. Mosteller subsequently worked on medical technology assessment with Henry Beecher and numerous other anesthesiologists at Massachusetts General Hospital in Boston.

Mosteller served on the President’s Commission on Federal Statistics (1971). His group, which included Richard Light and Herbert Winokur, looked at field studies in federal reports and recommended that field studies and randomized trials should be used more frequently in analyzing social, health, and welfare programs, and that the government should work on improving the climate for performing field studies.

Working with Daniel Patrick Moynihan, then a professor of government at Harvard, Mosteller co-wrote the influential and thought-provoking work On Equality of Educational Opportunity (1972). In response to the polemical Coleman Report on inequalities in public education, Moynihan and Thomas Pettigrew organized a large university-wide seminar at Harvard to debate the report. Mosteller became involved in the seminars, played a major role in the statistical reanalysis of the data from the report, and partnered with Moynihan in editing and organizing the written work that evolved from the discussions.

Frederick Mosteller believed in the importance of statisticians taking part in empirical work, and would often encourage his students and colleagues to broaden the scope of their studies. This belief enabled his involvement with the School of Public Health at Harvard in 1973. Mosteller and Howard Hiatt, dean of the school, established an interdisciplinary faculty seminar on health and medicine. The seminar addressed another important need. Prior to the formation of the seminar, little work had been done on evaluating procedures in health and medicine. Smaller groups also grew out of the seminar. Mosteller chaired the Surgery Group, and continued his involvement with this group into the 1980s. At least one major work came from the seminar. Working with fellow seminar participants John Bunker and John Barnes, Mosteller edited an influential work on the subject of evaluating medical procedures-- Costs, Risks, and Benefits of Surgery (1977).

Influenced by the success of the faculty seminar at the School of Public Health, Howard Hiatt sought to strengthen the quantitative side of the health services at Harvard. With the approval of the retiring chairman of the Biostatistics department, Hiatt approached Mosteller in 1977 to lead the department. Given the opportunity to expand the department and to pursue research in medical statistics, Mosteller was happy to accept. Under his leadership, the biostatistics department expanded and prospered. Among those added to the staff by Mosteller were noted biostatistician Marvin Zelen and a group of statisticians who were experts in clinical trials, particularly in trials dealing with cancer. During his time at Biostatistics, Mosteller, along with Lawrence Thibodeau and James Ware, also developed a quantitative course in biostatistics for physicians.

After four years heading Biostatistics, Mosteller was approached to lead another department in the School of Public Health, the department of Health Policy and Management (HP & M). The department was in its embryonic stages of development and was comprised of heterogeneous disciplines. It included faculty with business training and specialists in health care, economists, evaluators, and policy analysts. HP & M was in need of a new organization and a new leader, an experienced administrator with a strong research background and an interest in policy. Prior to Mosteller’s arrival, the department was disorganized, loosely grouped around a series of committees. Under Mosteller’s influence, a new direction for HP & M began to take shape. He was able to reorganize the administration and to promote a team-first attitude among a disparate faculty by instituting regular department colloquia. Additionally, he strengthened the academics of the department by promoting a strong doctoral program and stressing the importance of international health. HP & M thrived under Mosteller’s guidance for six years (1981-1987).

In addition to his departmental and teaching duties, Mosteller continued to be involved with policy formulation and interdisciplinary pursuits. He chaired the Health Science Policy Working Group, which met monthly and attracted university-wide interest among several schools and departments. In 1987, he also became involved with the Technology Assessment Group, which focused on evaluating medical technologies. A primary emphasis of this group was meta-analysis, the combination and analysis of quantitative information from several comparable studies, and it provided a useful look in to the complex process by which advances in medical technology affect policies that promote the public good.

In 1990 Howard Hiatt tapped Mosteller to lead a project entitled “Initiatives for Children” at the American Academy of Arts and Sciences. The initiatives sought to provide an analysis of programs affecting children, particularly educational programs, from birth through college. The scope of the project was quite broad. In addition to education, it also addressed areas of health, discipline, law, and public policy. The project began as a series of seminars on education, and eventually led to several studies and publications.

In addition to his contributions to academe and public policy, Mosteller was one of the first to apply serious statistical modeling to sports. The loss of his favorite baseball team, the Boston Red Sox, to the St. Louis Cardinals in the 1946 World Series sparked a lifelong interest in sports statistics. His subsequent work published in 1952, analyzing the probability of the better team winning a World Series, was particularly ground-breaking. Throughout his career, he undertook further sports-related studies on baseball, football, and golf, and was fond of using sports examples in his textbooks. Besides enjoying sports, other hobbies of Mosteller included magic, playing cards, and spending time at his summer home on Cape Cod.

After officially retiring from teaching and administration in 1992, Mosteller continued to play an active role in the world of statistics. He maintained an office in the Department of Statistics at Harvard, undertaking research, writing, and advising other scholars. His papers attest to the breadth of scholarship which he continued to pursue, up until the final chapters of his life. Mosteller’s wife, Virginia, predeceased him, passing away in 2001. Frederick Mosteller died in Arlington, Virginia, after a long illness on July 23, 2006.

Mosteller was widely decorated and his scholarship and accomplishments were recognized by numerous academic and honorary institutions. He received honorary doctorate degrees from the University of Chicago (1973), Carnegie Mellon University (1974), Yale University (1981), Wesleyan University (1983), and Harvard University (1991). Some of the honorary societies he belonged to included: American Academy of Arts and Sciences, American Philosophical Society (1961), Institute of Medicine, International Statistical Institute, and the Royal Statistical Society. He served as the president of several professional bodies including 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. In 1992, the Frederick Mosteller Professorship was established at the Harvard School of Public Health.

Frederick Mosteller has the distinction of being the only person to have chaired four different departments at Harvard: Social Relations (1953-1954); Statistics (1957-1969, 1973, 1975-1977); Biostatistics (1977-1981); and, Health Policy and Management (1981-1987).