American Scientist

THE PLEASURES OF STATISTICS: The Autobiography of Frederick Mosteller. Frederick Mosteller. Edited by Stephen E. Fienberg, David C. Hoaglin and Judith M. Tanur. xvi + 344 pp. Springer, 2010. $39.95.

The career of Frederick Mosteller exhibits superbly the disarming modesty of statistics. Mathematicians to some degree look down on statistical knowledge for its makeshift character, and although we can expect an educated person to know of the leading personages in physics, biology and economics, few outside the quantitative sciences would be able to name a single statistician. Yet statistics, as the master science of numbers and measures, has the standing to organize the efforts of professionals (physicians and policy experts, for example) who have greater wealth and status than do statisticians. Statistics is able to do this all the more effectively because it so rarely calls attention to itself as a creative activity, or to the numbers it analyzes and often generates as products of negotiated expertise and discernment.

Born in 1916, Mosteller was raised in modest circumstances, if not real poverty, near Pittsburgh. If he ever dreamed of greatness or mapped out his road to success, he omits all this from his chapters about growing up. His success in high school mathematics and particularly in a state scholarship examination gave him the resources to attend the Carnegie Institute of Technology, as it then was, where he pursued studies and made contacts that facilitated his move to Princeton University as a graduate student. In 1944–45 he took part in war work with the Statistical Research Group at Columbia University, an enterprise whose significance for American statistics might be compared with that of Los Alamos for nuclear physics. Institutionally, statistics had just begun to take shape as a discipline, though already it was becoming indispensable to public health, medicine, medical policy, business, engineering, government and the sciences, especially the social sciences.

Upon completing his doctorate in 1946, Mosteller joined the new Department of Social Relations at Harvard, created by the sociologist Talcott Parsons in an effort to advance interdisciplinary social research. Parsons had worked to establish a basis in theory for the unity of the social (or behavioral) sciences, but he was committed also to empirical research, and this, more and more, meant quantification. Mosteller, the poor boy from Pittsburgh, thus assumed a central position in a key academic center at a moment of unmatched promise for social science. This was in 1946, the year he turned 30, although in photographs he appears much older. He evinces in his autobiography not a hint of wonder at his meteoric ascent, nor a moment of self-doubt in the course of supervising a multitude of dissertations, directing centers, founding and chairing departments, and serving on expert commissions.

Mathematics, like romantic poetry, has become associated in modern times with moments of revelation and of genius. That is not Mosteller’s way. Recognizing that readers must sooner or later begin to wonder how he could have taught so many students, written so many books, and organized and executed such diverse projects with collaborators from so many disciplines, he responds with advice on making every moment count. He does not think of wrestling with imponderable questions or of asking what it all means. Statistical science, the keystone in the arch of public rationality, seems to advance systematically in little steps, and so it is also in his working life. The briefest idle moment, he tells us, provides time to write a paragraph or comment on a manuscript, provided he can react to each fragment separately rather than waiting until he reaches the end and assessing the work as a whole.

Mosteller believed in statistics as a proper discipline, and he describes here how he won support to create a department at Harvard. But statistics was always for him a vital tool of scientific and social research. He comes across as a man of genuine if sometimes strategic kindness who knows how to win friends and does all he can to avoid making enemies. Again and again his role as expert statistician provided the occasion to exercise these powers. By his own account, he spent the greater part of his career assisting in research involving social and economic policy or the assessment of medical effectiveness. Proper statistical analysis may give the appearance of a certain narrowness, yet statistics often takes on a key role in the overall design of surveys and experiments. The statistician, while perhaps knowing rather little about the parts, may yet have a uniquely privileged view of the whole. Mosteller believed deeply in such research as the proper mission of statistics. He worked assiduously as a teacher and as an organizer of interdisciplinary seminars to assure that statistics should not retreat into disciplinary isolation but assume an active role in the world, with the goal of enhancing efficiency. Yet he was never so blind as to let a single bottom line, such as monetary gain or lives saved, define the worth of an outcome. In medicine, for example, he did not favor adding a few weeks to an average life at the cost of months of expensive treatment with attendant suffering of the patient for whose sake the effort is nominally undertaken.

In other ways, though, Mosteller’s vision seems too narrowly focused. He acknowledges that statistical studies are often bound up with legal, commercial and administrative processes. The first of the “examples of quantitative studies” with which he begins the book concerns the embarrassing prediction by all pollsters that Harry Truman would be defeated in the presidential election of 1948. An expert commission of inquiry, on which Mosteller served, was quickly assembled to find out why the polls had all been wrong. The leaders of social science were concerned in part because so conspicuous a failure might undermine the faith of marketing departments in the objectivity of sample surveys. Mosteller, however, says nothing about corporate sponsorship of clinical trials, or about the inevitable problems of missing data and other ambiguities that make these studies so susceptible to corruption when powerful interests stand to gain by perverting them. That is, the picture of statistics he presents here is one that emphasizes scientific concepts and designs, and tends to downplay both the interests at stake and the close-to-the-ground labor through which these designs are, always imperfectly, enacted.

Like most scientists, Mosteller is reluctant merely to put the focus on his life and career, wanting instead to make his memoir a paean to and lesson in statistics. The difficulty of reconciling biography with science seems to have occasioned for him serious and highly uncharacteristic problems of writing. His guiding principle, as he and the book’s editors all tell us, was to work first at the task that is nearest completion. Yet somehow his autobiography languished over two decades and remained incomplete at his death. Fully a third of the published manuscript, up to page 110, consists of potted stories of statistics in action, some familiar from other popular writings. In the biographical sections he tells of his family background but virtually omits his wife and children, except as they figure in a chapter on their summer home in Cape Cod, where windows must be repaired and raccoons kept away from trash cans. The logic of his story requires the Cape to be mainly a place of scientific work. Little enthusiasms—specifically, bridge, chess and magic—can come into the book because they seem vaguely pertinent to statistics. Mosteller announces his strong preference for writing in the plain style and comments that statistics has solved the problem of metaphor, which he nevertheless prefers to avoid.

In short, in this autobiography he was not able to reconcile his divergent purposes or to achieve a harmony of style. Yet the thoroughly flawed book provides a fascinating view of statistics, not so much when Mosteller sets out to give statistical lessons as when, almost inadvertently, the personage of the statistician stands revealed by this narrative of a life devoted to quantitative reason.

Theodore M. Porter is a professor in the Department of History at the University of California, Los Angeles. His books include Trust in Numbers: The Pursuit of Objectivity in Science and Public Life (Princeton University Press, 1995) and Karl Pearson: The Scientific Life in a Statistical Age (Princeton University Press, 2004).