A gleaning for my colleagues from Sunday's New York Times. This is the
entire article.
Gordon Bear School of Science Ramapo College Mahwah NJ 07430-1680 USA
-----------------------------------------------------------------------------
Voice: 201 684-7754 Fax: 201 684-7637 E-Mail: [EMAIL PROTECTED]
Leslie Kish, 90; Improved Science of Surveys
October 14, 2000
By ERIC PACE
Leslie Kish, an authority on scientific population sampling who
predicted President Harry S. Truman's upset victory over Thomas E.
Dewey in 1948, died on Oct. 7 in a hospital in Ann Arbor, Mich. He
was 90 and lived in Ann Arbor.
Professor Kish and the influence of his teachings "fundamentally
changed how the world collects information about itself," said
Robert M. Groves, a professor of survey methodology at the
Universities of Michigan and Maryland, and a former student of
Professor Kish, who taught at the University of Michigan.
Professor Kish developed sampling methods that are employed now in
censuses as well as in political surveys and polls. The basic task
of sampling is to draw a representative sample from a universe of
possible observations.
"He took a set of theoretical developments in statistics that
occurred in the late 1930's and translated them to applications in
hundreds of social science research projects throughout the world,"
Professor Groves said.
Another former student, Martin R. Frankel, a professor of
statistics and computer information systems at Baruch College of
the City University of New York, said Professor Kish had found
applications for the new ideas in statistics "in fields where
surveys were emerging as a way of looking at things." As examples
he mentioned economic behavior surveys and a study of the use of
health insurance.
In the 1930's most pollsters used quota sampling, in which
interviewers are asked to find and question certain numbers of
people who match various profiles. But a small group of American
statisticians was beginning to use probability sampling, in which
interviewees are chosen by a random process outside the
interviewer's control. By the 1940's there was debate on which
method should become the standard.
Professor Kish was instrumental in getting probability sampling
recognized, at Michigan and elsewhere, as the type of sampling to
be used in surveys involving human behavior.
The value of the probability sampling methods that Professor Kish
worked out was shown in the 1948 presidential election.
He and colleagues at Michigan carried out a probability sample of
fewer than 1,000 American households. At the time, news media and
commercial polls were saying that Dewey would win easily. But the
Michigan sample indicated that Truman was slightly ahead.
Truman won. Probability sampling is now the accepted method of
scientific sampling.
Professor Kish was also an early advocate of rolling samples. That
is, the repeated use of separate sample surveys a different
sample each time at regular intervals, that cumulate into very
large studies. That technique is is just beginning to be used in
the United States.
In addition, Professor Kish wrote "Survey Sampling" (1965, Wiley),
which is still in print and widely used.
He was on the Michigan faculty from 1951 until his formal
retirement, as a professor of sociology, in 1981. He was a
co-founder in 1947 of the Institute for Social Research at the
university.
Professor Kish was born in Poprad in what is now Slovakia. He came
to the United States with his family in 1926, settled in the Bronx
and went on to receive a bachelor's degree from the City College of
New York and a master's degree and his doctorate, both from
Michigan.
In 1937 he joined the International Brigade to fight Franco in the
Spanish Civil War. He saw action in a Hungarian battalion was
wounded and returned to the United States in 1939.
He is survived by his wife, the former Rhea Kuleske; two
daughters, Carla Kish of Silver Spring, Md., and Andrea Kish of St.
Paul; and a sister, Magda Bondy of White Plains.
Copyright 2000 The New York Times Company
=================================================================
Instructions for joining and leaving this list and remarks about
the problem of INAPPROPRIATE MESSAGES are available at
http://jse.stat.ncsu.edu/
=================================================================
