|
|
William G. Cochran
1909-1980
|
William Gemmell Cochran was born on July 15, 1909, in Rutherglen, Scotland,
arguably the oldest royal burgh in the country. Cochran and his one brother
had a typical childhood for a middle-class family, and Cochran did well in school
winning many prizes for academics. This set the stage for his higher education
experience; first place in the University of Glasgow Bursary Competition placed
him in that school for his first degree, and the Logan Medal for the best Faculty
of Arts student and a scholarship put him at Cambridge to study mathematics,
applied mathematics, and statistics under John Wishart.
Cochran began his career at Rothamsted Research without a PhD, a novelty for
the firm. Yates was the one who persuaded Cochran to take the job, believing
there was little left for him to learn in school about research that he couldn't
pick up from actually conducting research. Cochran published 18 papers while
at Rothamsted and attended the lectures of R.A. Fisher before leaving England
for the United States. He stayed for six years, experimenting with weather issues,
differential fertility of plots, and lack of replication, and working with Yates
on sample surveys. The practical value of Cochran's work is shown in the understanding
of the importance of randomization in sample surveys and experiments. He found
that people picking wheat shoots tended to choose those with an average height
exceeding the field average, but avoided shoots that were overly tall or short.
Cochran published his conclusions in a 1936 paper in the Empire Journal of
Experimental Agriculture.
In 1939, Cochran began teaching statistics at Iowa State University, working
with Raymond Jessen on sampling developments. He was tasked with developing
the graduate program in statistics within the Mathematics Department. During
this time, Cochran also worked on the advisory panel to the U.S. Census, serving
as chair.
Cochran worked with the Statistical Research Group and Samuel Wilks at Princeton
from 1943 to 1944. The team studied naval warfare, examining the probabilities
of hits. In 1945, Cochran was developing bombing raid strategies.
After World War II, Cochran moved to North Carolina to help found the graduate
program at the North Carolina Institute of Statistics, which spans several
universities. Three years later, he moved on to chair Johns Hopkins University's
Department of Biostatistics, where his work shifted from agricultural issues
to medical applications of statistics. While at Johns Hopkins, he wrote Sampling
Techniques and Experimental Designs; the latter was coauthored with Gertrude
Cox. From 1957 until his retirement in 1976, Cochran worked at Harvard. He
was hired to help set up Harvard's statistics department, a task for which
he was well-prepared, and his last position was Professor Emeritus.
Throughout his career, Cochran played a role in the field outside the university,
working on the Kinsey Report on human sexual behavior, the effects of radiation
on Hiroshima victims, the search for a polio vaccine, surgical experiments for
ulcers, and educational opportunity equality. He is especially well-known for
researching the effects of smoking for the U.S. Public Health Service. In 1964,
based on the work of Cochran, and other scientists, the Advisory Committee presented
the Surgeon General with proof that lung cancer was a direct result of smoking.
This was the first official recognition of the health risks connected to tobacco.
Cochran discovered many statistical methods, such as those for including or
excluding an independent variable in multiple linear regression. The Cochran
Q-Test is used to evaluate two variables measured on a nominal scale. He is
remembered primarily for his agricultural studies, such as the Influence of
Rainfall on the Yield of Cereals and the Field Counts of Diseased Plants. Cochran
participated in the National Academy of Sciences-National Research Council panel
on incomplete data in simple surveys from 1974 until his death in 1980. Service
to the profession did not end with Cochran's contributions to the practice of
statistics. He was President of the Institute of Mathematical Statistics in
1946, succeeding W. Edwards Deming for the position. In 1953, he served as the
48th President of the ASA, then as President of the International Biometric
Society from 1954 to 1955. In 1959, Cochran was elected honorary fellow of the
Royal Statistical Society; he held a Guggenheim Fellowship in 1964 and won the
S.S. Wilks Medal of the American Statistical Association in 1967 "for his
many contributions to the advancement of the design and analysis of experiments
and their value for military research." Cochran was awarded honorary doctorate
degrees from the University of Glasgow and Johns Hopkins University.
Cochran's teaching skills lived on in his 40 or more PhD supervisees, many
of whom went on to become leading statisticians. As a teacher, he was known
for clarity and individuality, as well as his high expectations for his students.
He was always available to his students, his colleagues, and the public, and
was often sought out for help on a stubborn problem. When he retired in 1976,
hundreds of admirers from all over the world traveled to the Harvard Club in
Boston to wish him well.
Related Links
http://www-groups.dcs.st-and.ac.uk/~history/Mathematicians/Cochran.html
