R.A. Fisher: An Appreciation

218 Pages·1980·3.48 MB·English

by  David Hinkley (auth.)

L_ecture Notes in
~)tatistics
Edited by J. Berger, S. Fienberg, J. Gani,
K. Krickeberg, I. Olkin, and B. Singer
1
E•. E. Fienberg D. V. Hinkley
(Eds.)
FtA. Fisher:
Appreciation
P~n
Springer-Verlag
Berlin Heidelberg New York London Paris Tokyo Hong Kong
Editors
Stephen E. Fienberg
Department of Statistics, Carnegie Mellon University
Pittsburgh, PA 15213, USA
David V. Hinkley
Department of Statistics, Oxford University
1 South Parks Road, Oxford OX1 3TG, United Kingdom
1st Edition 1980
2nd Printing 1989
Mathematical Subject Classification: 01 A60, 01 A 70, 62-00, 62-03, 62Cxx,
62D05, 62Exx, 62Fxx
ISBN-13: 978-().387-004764 &lSBN-13: 978-1-4612.6)79.Q
001: 10.1007l978-1-4612.6)79.Q
This work is subject to copyright. All rights are reserved, whether the whole or part of the material
is concerned, specifically the rights of translation, reprinting, re-use of illustrations, recitation,
broadcasting, reproduction on microfilms or in other ways, and storage in data banks. Duplication
of this publication or parts thereof is only permitted under the provisions ofthe German Copyright
Law of September 9, 1965, in its version of June 24, 1985, and a copyright fee must always be
paid. Violations fall under the prosecution act of the German Copyright Law.
© Springer-Verlag Berlin Heidelberg 1980
Printing and binding: Druckhaus Beltz, Hemsbach/Bergstr.
2847/3140-543210 - Printed on acid-free paper
Steward at First International Eugenics Congress. 1912
Fellow of the Royal Society, March 1929
Fisher at his desk calculator at Whittingehame Lodge, 1952
Preface for the Second Printing
This year, 1990, marks the 100th anniversary of the birth of Sir Ronald A.
Fisher, arguably the most important and most influential statistician of the
century and perhaps in the history of the field. Thus we are pleased that
Springer-Verlag has agreed to reprint this volume in honor of the occasion.
When this c::>liection of essays was published a decade ago as the first
volume in the Springer Lecture Notes Series in Statistics, we were unsure of
just who would be interested in the material developed in our faculty-student
seminar at the University of Minnesota. Several thousand statistical colleagues
was the answer to the question implicit in our uncertainty. We ourselves have
come back to these papers on a number of occasions in the course of our own
research activities.
Several of the authors of papers in this volume have since relocated,
including the editors, and we have chosen not to revise the affiliations listed
in the Table of Contents. We do wish to take special note regarding two authors
for whom we had special affection and who have died during the intervening
decade -- Wi"iarn Cochran and Robert Buehler.
Stephen E. Fienberg
David V. Hinkley
A very large number of the now-standard techniques of modern statistical theory
and methods have their origins in the work of Sir Ronald A. Fisher. Yet our full
professional debt to this great scientist often goes unacknowledged in modern
courses on theoretical and applied statistics, and in modern textbooks. We believe
that it is profitable for a student to consider the origins of modern-day statistics,
in part because the original motivations for particular lines of development and
philosophies have become clouded by time. This is particulary true of Fisher's
work, which to an unfortunate extent has been superceded in the literature by work
of a more mathematical nature. To study Fisher's statistical writings is to become
enlightened, surprised -- and perhaps sometimes infuriated. To attempt to under
stand Fisher, and to argue over his writings, is to attempt a fundamental under
standing the main threads of the fabric of statistical inference, and of modern
statistical methods.
The present volume of lectures contains edited versions of presentations
delivered as part of a faculty-student seminar and during a Special Lecture Series
in the School of Statistics, University of Minnesota, during the Spring of 1978.
The seminar was a cooperative effort, involving almost all of the faculty and
graduate students in the School, and focussed on some of Fisher's major papers in
the fields of estimation and statistical inference, design and analysis of
experiments, contingency tables, distribution theory, discriminant analysis, and
angular statistics. These papers are concentrated primarily in a 15 year period,
1920-1935, which saw giant strides in our subject and great debates involving
Fisher and his contemporaries. The lectures were not intended as authoritative
critiques of Fisher's work, but rather as introductions to and reviews of the key
ideas in his writings. The same is true of the edited versions included here. The
ordering of the lectures follows in a rough way the chronological ordering of the
material being discussed.
-ix-
In conjunction with the seminar the School of Statistics, with the support of
the University of Minnesota Department of Concerts and Lectur,es and the College
of Education, sponsored a Special Lecture Series llnder the general title: "R.A.
Fisher: An Appre,:iation", now the title of this volume. Each lecture was given by
a distinguished vLsitor, and three of these, by Joan Fisher Box, William Cochran,
and David Wallace. fitted conveniently into the outline of material covered by
the regular seminar series. Thus edited versions of these three lectures are
included in the PJ~esent volume. The remaining Special Lectures included a presen
tation on pre-Fisherian work by Stephen Stigler, and discussions of work that has
built on Fisher's, by D.A.S. Fraser, I.J. Good, Marvin Kastenbaum, and Oscar Kemp
thorne.
Not only was Fisher one of the most productive and original statisticians of
this century, but he also made fundamental contributions to genetics. There is an
apocryphal story of a geneticist who, when introduced to a statistician several
years ago, commented: "There's a well-known geneticist who I am told dabbles
occasionally in statistics. Perhaps you have heard of him -- his name is Fisher".
Actually, the extEnt of Fisher's writing in genetics (as well as related material
in eugenics) is remarkable. One need only scan the complete list of Fisher's
publications, reproduced here as an Appendix, to be impressed. While this material
is clearly worthy of study, it lies beyond the boundaries of the seminar and Lecture
Series, and is not discussed in the present volume.
Because so many of Fisher's publications are referred to in several lectures,
we have reproduced his entire list of publications, as given in the Collected
Pap~_s_~±-L:i.';her edited by J.H. Bennett (The University of Adelaide, South
Australia: Coudrey Offset Press, 1974), at the end of this volume. References to
Fisher's papers iIi the text are given in the style, CP 48, referring to paper No.
48 in the Collected Papers. All papers and books by others referenced in the
lectures are list€~d in standard form at the end of each presentation.
The present volume of lectures has grown out of a detailed set of notes pre
pared by Christy Chuang, and typed by Karen Bosch. All of the lectures were
-_x-
subsequently rewritten, edited, and retyped with extensive assistance from Linda
D. Anderson, who also prepared the final typescript. Without her skills and efforts
the volume would not have been completed. The figures were drawn by Henri Drews and
Sylvia Giesbrecht of the Department of Information and Agricultural Journalism at
the University of Minnesota. Joan Fisher Box kindly supplied the photographs used.
This work was supported in part by Office of Naval Research Contract N00014-78-C-
0600 and National Science Foundation Grant MCS 7904558 to the Department of Applied
Statistics, School of Statistics, University of Minnesota. Reproduction in whole
or in part is permitted for any purposes of the United States Government.
We hope that the lectures here will help to introduce others to Fisher's
work, so that yet another generation of statisticians can gain an appreciation of
the relevant impact that his ideas still llave on the way scientific research is
conceived, carried out, and understood.
SEF
DVH
Minneapolis and St. Paul, Minnesota
November, 1979
CONTENTS
R.A. FISHER: SOME INTRODUCTORY REMARKS
David V. Hinkl"y, School of Statistics, Twin Cities, Minnesota 1
FISHER: THE EARLY YEARS
Joan Fisher Box, Madison, Wisconsin 6
DISTRIBUTION OF THE CORRELATION COEFFICIENT
Somesh Das Gupta, School of Statistics, Twin Cities, Minnesota 9
FISHER AND THE ANALYSIS OF VARIANCE
William G. Cochran, Professor Emeritus, Harvard University 17
RANDOMIZATION AND DESIGN: I
Norton Holschuh, School of Statistics, Twin Cities, Minnesota 35
RANDOMIZATION AND DESIGN: II
Rick Picard, Sehool of Statistics, Twin Cities, Minnesota 46
BASIC THEORY OF THE 1922 MATHEMATICAL STATISTICS PAPER
Seymour Geisser, School of Statistics, Twin Cities, Minnesota 59
FISHER AND THE METHOD OF MOMENTS
Roy Mensch, School of Statistics, Twin Cities, Minnesota 67
FISHER'S CONTRIBUTIONS TO THE ANALYSIS OF CATEGORICAL DATA
Stephen E. Fienberg, School of Statistics, Twin Cities, Minnesota 75
THEORY OF STATISr::CAL ESTHIATION: THE 1925 PAPER
David V. Hinkley, School of Statistics, Twin Cities, Minnesota 85
SOME NUMERICAL ILLUSTRATIONS OF FISHER'S THEORY OF STATISTICAL ESTIMATION
George Runger, School of Statistics, Twin Cities, Minnesota 95
FISHER'S DEVELOPMENT OF CONDITIONAL INFERENCE
David V. Hinkley, School of Statistics, Twin Cities, Minnesota 101
FIDUCIAL INFERENCE
Robert Buehler" School of Statistics, Twin Cities, Minnesota 109
THE BEHRENS-FISHEll AND FIELLER-CREASY PROBLEMS
David L. Wallaee, Department of Statistics, University of Chicago 119
FISHER, JEFFREYS, AND THE NATURE OF PROBABILITY
David A. Lane, School of Statistics, Twin Cities, Minnesota 148
DISCRIIIINANT ANAL"SIS
Somesh Das Gupta, School of Statistics, Twin Cities, Minnesota 161
DISTRIBUTION ON TilE SPHERE
Christopher Bingham, School of Statistics, Twin Cities, Minnesota 171
SMOKING AND LUNG CANCER
R. Dennis Cook, School of Statistics, Twin Cities, Minnesota 182
APPENDIX: PUBl,ICATIONS OF R.A. FISHER 192