One purpose of this International Year of Statistics is to spread the word that the field of statistics benefits society. As part of the International Year, many organizations, including SAS and the American Statistical Association (ASA), are turning to history to illustrate how statistics is vital to the health and welfare of the world.
Recently, Stephen Stigler wrote an article in the ASA membership magazine in which he singled out 20 ASA members who were "influential in bringing [the ASA] to this point in our history." Stigler readily admits that "many excellent people have been omitted" from the list. I would like to highlight one of those who were omitted. This is the biography of Jerome Cornfield, who not only served as a president of the ASA, but also made fundamental contributions to the fields of statistics, medical research, and epidemiology.
Jerome Cornfield: The early years
Jerome Cornfield was born the son of Russian Jewish immigrants in 1912. Like John Sall, co-founder of SAS, he got a degree in history. He joined the US Bureau of Labor Statistics during the Depression (1935) and learned statistics by taking courses at the US Department of Agriculture from 1936–1938 (Greenhouse, p. 3). During this time he used probability sampling methods (which were cutting-edge research in the 1930s) to determine the number of unemployed in the US. He was instrumental in the design of samples used by the federal government, and contributed to the revision of the Consumer Price Index (1938-1940). In 1947 he joined a group that provided consultation to the fledgling National Institutes of Health (NIH) (Greenhouse, p. 4).
The link between cigarettes and lung cancer
At the NIH, he became interested in a vexing problem: Why were so many people dying of lung cancer in 1950, when 50 years earlier there were so few cases reported? Was it better diagnosis? More factory pollution? A population that was living longer than ever?
Some retrospective case-control studies by Richard Doll and Bradford Hill in the UK had linked lung cancer to cigarette smoking. But Cornfield wanted a stronger conclusion. He wanted to know whether the correlation was actually a cause-and-effect relationship. He also wanted to use statistics to assess the risk that an individual who smokes would develop lung cancer. By using Bayes' rule, Cornfield was able to combine Doll and Hill's results (namely, the estimated probability that someone was a smoker, given that they had lung cancer) with NIH data to answer the inverse question: the probability that someone would develop lung cancer, given that he was a smoker. The result: smokers are many times more likely to develop lung cancer than nonsmokers.
Cornfield's work "stunned research epidemiologists" and his techniques "made much of modern epidemiology possible" (McGrayne, p. 111). Although his work provoked the ire of the volatile and influential Sir Ronald Fisher (Fisher was a smoker, a paid consultant to the tobacco industry, and a fervent anti-Bayesian), Cornfield successfully defended his methods and results. In 1964, when the US Surgeon General warned "that cigarette smoking is causally related to lung cancer" (McGrayne, p. 113), Cornfield's work was among the evidence that was cited.
The risk factors for heart disease
Cornfield went on to apply statistics to many other controversial and important problems, such as the safety of the polio vaccine, and the safety and efficacy of drugs and food additives (Greenfield, p. 4). Among his achievements was contributing to the design of the Framingham Heart Study, a longitudinal study of the health of residents of Framingham, Massachusetts. Between 1948–1958, 92 out of 1,329 Framingham males experienced some form of a cardiovascular event, such as a heart attack or stroke. From these results, Cornfield was able to use statistics to identify age, cholesterol, cigarette smoking, heart abnormalities, and blood pressure as major risk factors that contribute to cardiovascular disease (McGrayne, p. 115). The identification of these risk factors is responsible for one of the most important public health achievements of the 20th century: the dramatic decline of mortality due to cardiovascular disease.
Contributions to the ASA
Besides serving as the ASA president in 1974, he was the president or vice-president of numerous other professional societies. He served as an editor of six scholarly journals, including the influential Journal of the American Statistical Association. His presidential address to the ASA is an amusing essay entitled "A Statistician's Apology," in which he asks "why should [a statistician] of spirit, of ambition, of high intellectual standards, take any pride or receive any real stimulation and satisfaction from serving an auxiliary role on someone else's problem?" (Cornfield, p. 9) The answer, he asserts after given examples from his own career, is that there is substantial "stimulation and satisfaction" from interacting and collaborating with scientists in other fields.
Furthermore, he continues, it is only by working on practical problems that statistics advances. "Application requires understanding, and the search for understanding often leads to, and cannot be distinguished from, research. The true Joy is to see the breadth of application and breadth of understanding grow together" (Cornfield, p. 11).
In this International Year of Statistics, it is important to remember the contributions of prolific researchers such as Jerome Cornfield. Yet, Cornfield's true legacy is that today's statisticians continue to use the methods that he introduced to understand the connection between risk factors and disease. Thanks to Cornfield and thousands of other statisticians, today's world is safer and healthier than ever before.
Cornfield, Jerome (1975). "A Statistician's Apology," Journal of the American Statistical Association, 70(349), pp. 7–14.
Greenhouse, Samuel W. (1982). "A Tribute," Biometrics, Proceedings of "Current Topics in Biostatistics and Epidemiology." A Memorial Symposium in Honor of Jerome Cornfield, 38, pp. 3–6
McGrayne, Sharon B. (2011). The Theory That Would Not Die, Yale University Press, Chapter 8, pp. 108–118.