The Work of Raymond J.

Carroll The Impact and Influence of

a Statistician

Visit the link below to download the full version of this book:

https://medipdf.com/product/the-work-of-raymond-j-carroll-the-impact-and-influen
ce-of-a-statistician/

Click Download Now

Editors
Marie Davidian Xihong Lin
Department of Statistics Department of Biostatistics
North Carolina State University Harvard School of Public Health
Raleigh, NC, USA Boston, MA, USA

Jeffrey S. Morris Leonard A. Stefanski

Department of Biostatistics Department of Statistics
The University of Texas North Carolina State University
MD Anderson Cancer Center Raleigh, NC, USA
Houston, TX, USA

ISBN 978-3-319-05800-9 ISBN 978-3-319-05801-6 (eBook)

DOI 10.1007/978-3-319-05801-6
Springer Cham Heidelberg New York Dordrecht London
Library of Congress Control Number: 2014939953

© Springer International Publishing Switzerland 2014

This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of
the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation,
broadcasting, reproduction on microfilms or in any other physical way, and transmission or information
storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology
now known or hereafter developed. Exempted from this legal reservation are brief excerpts in connection
with reviews or scholarly analysis or material supplied specifically for the purpose of being entered
and executed on a computer system, for exclusive use by the purchaser of the work. Duplication of
this publication or parts thereof is permitted only under the provisions of the Copyright Law of the
Publisher’s location, in its current version, and permission for use must always be obtained from Springer.
Permissions for use may be obtained through RightsLink at the Copyright Clearance Center. Violations
are liable to prosecution under the respective Copyright Law.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication
does not imply, even in the absence of a specific statement, that such names are exempt from the relevant
protective laws and regulations and therefore free for general use.
While the advice and information in this book are believed to be true and accurate at the date of
publication, neither the authors nor the editors nor the publisher can accept any legal responsibility for
any errors or omissions that may be made. The publisher makes no warranty, express or implied, with
respect to the material contained herein.

Printed on acid-free paper

Springer is part of Springer Science+Business Media (www.springer.com)

To Ray
Preface

Raymond J. Carroll’s impact on statistics and numerous other fields of science is far-
reaching and substantial. His vast catalog of work spans the spectrum from funda-
mental contributions to statistical theory to innovative methodological development
to new insights in a number of subject matter areas. From the outset of his career,
rather than taking the “safe” route of pursuing incremental advances, Ray has fo-
cused on tackling the most important statistical research challenges of our time, and
in doing so it is fair to say that he has literally shaped and defined a host of areas of
statistics, including weighting and transformation in regression, measurement error
modeling, quantitative methods for nutritional epidemiology, and non- and semi-
parametric regression. It is indisputable that Ray is one of the giants of the field, and
we are honored to have had the opportunity to prepare this volume, which highlights
some of his most influential work.
The book is organized into seven main parts, each focused on a key area in which
Ray has made significant contributions. The seven subject areas reviewed in this
book were chosen by Ray himself, as were the articles representing each area. Each
part is focused around these key papers, and, for each, we asked distinguished re-
searchers in the area to provide a commentary giving insight into not only the sig-
nificance of the featured papers but also on Ray’s impact on the area more broadly.
The commentaries not only review Ray’s work, but they also are filled with his-
tory and anecdotes that reflect the fact that Ray is also a really nice guy! Indeed, as
former students and collaborators of Ray, we are pleased that the personality, gen-
erosity, friendship, and enthusiasm we know so well emerge throughout all of the
commentaries, whose authors have almost all had the pleasure of working with Ray
firsthand as we have. We are deeply grateful to these contributors, whose thoughtful,
insightful commentaries provide an inspiring roadmap to Ray’s achievements. Due
to their extraordinary efforts, this book is a fitting tribute to a scholar and educator
whose influence on not only science but also on the individual students, postdocs,
and junior colleagues he has mentored is legendary.
Our elation with the authors who contributed their insights into Ray’s work and
personality is tempered by the death of George Casella. George provides an enter-
taining overview of Ray’s work in a hodgepodge of “Other” areas. He was both a

ix
x Preface

close friend and colleague of Ray. We are grateful that George was able to contribute
his personal reflections before his passing.
Putting together this volume was made even easier by Ray himself, and we can-
not thank him enough. He provided us with extensive materials, including not only
the list of articles around which the book is focused but also a detailed narrative of
his own thoughts on his work, his biography, and other resources.
We would also like to acknowledge Jennifer Moy, a student at North Carolina
State University, whose assistance in preparing Ray’s complete bibliography was
invaluable.
At the beginning of each commentary, the articles included in this volume that
form the basis for the commentary are listed and are identified by acronyms in brack-
ets; for example, “MEM” for Measurement Error Models.” The second number in
brackets is the number of citations reported by Google Scholar at the time Ray com-
piled the list (2011).
Of course, a book devoted to the contributions of Raymond Carroll cannot possi-
bly provide a full accounting of his work. Despite approaching the start of his fifth
decade as a researcher, Ray has not slowed his pace one bit, and he continues to pro-
duce and inspire and mentor students and postdocs unabated. We fully expect to be
called upon to put together “Volume 2,” featuring still other areas Ray has already
influenced and forthcoming contributions in areas that have yet to be defined.

Raleigh, NC Marie Davidian

Boston, MA Xihong Lin
Houston, TX Jeffrey S. Morris
Raleigh, NC Leonard A. Stefanski
December 2012
Contents

1 Measurement Error : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : 1
By John P. Buonaccorsi and Aurore Delaigle
2 Transformation and Weighting : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : 155
By David Ruppert
3 Epidemiology : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : 195
By Laurence Freedman, Mitchell H. Gail, and Dale L. Preston
4 Nonparametric and Semiparametric Regression
for Independent Data : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : 293
By Hua Liang

5 Nonparametric and Semiparametric Regression

for Dependent Data : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : 371
By Yehua Li and Naisyin Wang
6 Robustness : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : 447
By Roger Koenker and Douglas Simpson
7 Other Work : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : 509
By George Casella

Bibliography of Raymond J. Carroll . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 561

xi
Biography of Raymond J. Carroll

Raymond J. Carroll is Distinguished Professor of Statistics, Nutrition, and Toxicol-

ogy at Texas A&M University, where he has been on the faculty since 1987. He
was the first statistician ever given a Method to Extend Research In Time (MERIT)
Award from the National Cancer Institute (NCI) of the National Institutes of Health
(NIH), receiving this honor for his seminal contributions to statistical methodology
and the impact of that methodology on public health. He is the principal investigator
of an NCI-funded Bioinformatics training program and is the founding director of
the Texas A&M Center for Statistical Bioinformatics. He is also the Director for the
Texas A&M Institute of Applied Mathematics and Computational Science (http://
iamcs.tamu.edu).
Raymond Carroll was born April 21, 1949 in Yokohama, Japan, into an Irish
Catholic military family, and he is the eldest of five siblings. His father, who spent
the Second World War in India and China, was transferred successively from Yoko-
hama to Nagoya, Japan, Washington DC, Wichita Falls, Texas, Ramstein, Germany,
Wichita Falls, Omaha, Nebraska, and Seoul, Korea, and finally retired from his last
assignment in Wichita Falls. He is married to Marcia Ory. A memorial tree with a
plaque honoring the memory of his parents Regina and Norman is situated in the
heart of the central campus a few feet southwest of “Sully,” a bronze statue of the
first president of Texas A&M University. Three other memorial trees are adjacent,
two honoring the memories of his father-in-law, mother-in-law, and brother-in-law,
and the other honoring the memory of Don Risner, a good friend and fishing guide
from North Texas. Raymond attended high schools in Germany, Texas, and Ne-
braska. He graduated from the University of Texas at Austin in 1971 with a BA in
mathematics and was especially influenced by courses in analysis and measure the-
ory given by E. W. Cheney and G. W. Stewart, respectively. He received his PhD
in Statistics from Purdue in 1974 under the direction of Shanti Gupta, with won-
derful advice from Leon Gleser. He has held positions at the University of North
Carolina at Chapel Hill and the University of Pennsylvania. He has published over
350 papers and given over 300 invited talks. The peripatetic nature of his childhood
has made him an avid traveler, a characteristic not shared by his siblings. Since his
first invitation to Australia in 1987, he has visited that country over 20 times, and he

xiii
xiv Biography of Raymond J. Carroll

has visited Germany, the site of two of his sabbaticals, nearly yearly since 1980. He
is in addition a bad golfer who takes mulligans liberally and a mediocre although
enthusiastic fly fisherman.
Dr. Carroll is one of the world’s foremost experts on problems of measurement
error, data transformation, and nonconstant variation, and more generally on sta-
tistical regression modeling. His work has found application in a broad variety of
fields, including marine biology, laboratory assay methods, econometrics, epidemi-
ology, molecular biology, and many others. He has served as Editor of Biometrics,
a journal of the International Biometric Society, and as Editor of the Journal of the
American Statistical Association (JASA) Theory and Methods section. He has won
many honors in the profession, including the two major research awards. The first
is the 1988 Committee of Presidents of Statistical Societies (COPSS) Presidents’
Award, given annually by five major statistical societies to the outstanding statisti-
cian under the age of 40. Secondly, he gave the COPSS Fisher Lecture at the 2002
Joint Statistical Meetings, an award given by these statistical societies in honor of a
senior statistician “whose research has influenced the theory and practice of statis-
tics.”
Carroll’s work is characterized by a combination of deep theoretical advances,
innovative methodological development, and close contact with science. His first
seminal contribution to statistical methodology was to create methods for the analy-
sis of data with nonconstant variation; these methods being the transform-both-sides
method for nonlinear regression (together with David Ruppert) and the variance
function estimation approach (with Marie Davidian), both still in wide use. This
work developed from two projects, one on marine fisheries where he worked with a
team investigating how to model and manage the menhaden fishery in the Atlantic,
and the other project involving immunoassays at Eli Lilly and Company. In the early
1990s, with the inspiration of his close friend Mitchell Gail, he developed a deep
interest in epidemiologic case–control studies that led to his receiving the George
W. Snedecor Award from COPSS in 1997 for work in this area (together with Bruce
Lindsay and Katherine Roeder). The span of his scientific work is amazing, includ-
ing among many others (a) modeling ozone exposure in Houston (the 1997 JASA
Applications Editor’s Invited Paper); (b) understanding the effects of diet on breast
cancer; and (c) discovering interactions between genes and the environment (with
Nilanjan Chatterjee and Yi-Hau Chen).
Carroll is no doubt most well known for his work in the area of nonlinear mea-
surement error modeling, with applications to nutritional and radiation epidemiol-
ogy. The body of seminal research is of such depth, and of such importance, that at
the International Biometric Conference in 2000 in Berkeley, Scott Zeger described
him as the “grandfather” of measurement error modeling. His 1995 book and 2006
second edition with David Ruppert, Len Stefanski, and Ciprian Crainiceanu is the
standard reference in the field. This work began with his landmark 1984 paper in
Biometrika on measurement error in the binary regression framework and has con-
tinued to the present. He was the first to suggest the use of likelihood methods in
the nonlinear measurement error context. Along with Len Stefanski, he developed
the theory for and coined the name for regression calibration, the most commonly
Biography of Raymond J. Carroll xv

used method in nutritional epidemiology. His 1987 paper with Stefanski developed
the method of conditional score function. His 1990 paper with Stefanski and his
1988 paper with Peter Hall on deconvolution established the theoretical basis show-
ing how difficult it really is to understand latent variable distributions: this result
provides the theoretical underpinnings for the semi-parametric approaches in mea-
surement error models that have become increasingly popular. The deconvolution
area has become of great importance and interest, and even 20 years later the papers
have led others into the area. Carroll continues to produce important ideas, and his
work continues to influence others, in such important problems as mixed models,
segmented regression, instrumental variables, and nonparametric regression. More
recently, he has written papers on reanalysis of important radiation epidemiology
studies to account for measurement error, both in Biometrics.
Carroll’s work on measurement error modeling is also one of the landmark works
in nutritional epidemiology. He helped design the NCI-AARP Diet and Health
Study, the first study to confirm a link between fat in diet and breast cancer. He
was the senior author on the first major biomarker study (the OPEN Study) to un-
derstand how well common instruments such as the food frequency questionnaire
actually measure diet. This study was funded because of the methodological devel-
opments done together in what is now a long collaboration with Laurence Freed-
man, Victor Kipnis, and Douglas Midthune suggesting that the heart of the problem
of null studies was the instruments themselves.
Dr. Carroll has worked with many researchers from around the world, but no
doubt his closest collaboration has been with David Ruppert, now of Cornell Univer-
sity. They were next door office neighbors at the University of North Carolina from
1977 to 1987, where they started their original collaboration, and they have written
over 45 papers in addition to 4 books. Other colleagues with whom he has written 10
or more papers include Mitchell Gail, Victor Kipnis, and Douglas Midthune of the
National Cancer Institute; Peter Hall of the University of Melbourne; Len Stefanski
of North Carolina State University; Laurence Freedman of the Gertner Institute in
Israel; Naisyin Wang of the University of Michigan; Joanne Lupton, Nancy Turner
and Robb Chapkin, nutritionists at Texas A&M; Xihong Lin of Harvard; and Bani
Mallick of Texas A&M.
More recently, Dr. Carroll has developed a deep interest in basic molecular cell
biology and how it relates to nutrition and colon carcinogenesis. His research grants
include as co-investigator Dr. Joanne Lupton (the endowed Professor of Human
Nutrition at Texas A&M) and Dr. Nancy Turner. This work includes papers both in
biology journals and in JASA and Biostatistics, with many more papers under devel-
opment. Carroll is involved to the point of generating his own biological hypotheses,
suggesting new ways of measurement, and providing support so that novel mea-
surements can be undertaking to understand molecular pathways. More recently,
this close work with biologists and electrical engineers has led to the establishment
of an NCI-funded training program in Biostatistics and Bioinformatics, for which
Dr. Carroll has been the principal investigator since 2001, and the program has re-
cently been renewed until 2016. The program is unique because it aims to train
xvi Biography of Raymond J. Carroll

statisticians and electrical engineers in biology and includes mentors from biologi-
cal fields.
Dr. Carroll is an inspirational teacher and a major innovator for the Department’s
teaching program. In the 1990s he introduced the use of the computer and class
projects into STAT 302, an undergraduate course aimed at life science students.
Similarly, since 2000, in STAT 651 he was the first non-distance education expert to
create a distance course, something now routine in the department. Dr. Carroll has
won a College of Science Teaching Award, and he has graduated 35 PhD students,
many of whom are leading figures in academia and industry. He has also been the
mentor to many faculty members around the USA, including many who are now
full professors, and he is legendary for his willingness to give advice and technical
assistance.
PhD Students of Raymond J. Carroll

Name Institution Year

Gordon Johnston University of North Carolina at Chapel Hill 1979

Paul Gallo University of North Carolina at Chapel Hill 1981
David Giltinan University of North Carolina at Chapel Hill 1983
Len Stefanski University of North Carolina at Chapel Hill 1983
Doug Simpson University of North Carolina at Chapel Hill 1985
Marie Davidian University of North Carolina at Chapel Hill 1986
Stena Kettl University of North Carolina at Chapel Hill 1987
Yin Yin University of North Carolina at Chapel Hill 1988
Lie Ju Hwang Texas A&M University 1990
Jungsywan Sepanski Texas A&M University 1991
Rick Landin Texas A&M University 1992
C. Y. Wang Texas A&M University 1993
Ron Knickerbocker Texas A&M University 1993
Bobby Gutierrez Texas A&M University 1995
Stephen Eckert Texas A&M University 1995
Jeff Maca Texas A&M University 1997
Christian Galindo Texas A&M University 1998
Steve Iturria Texas A&M University 1998
Jeffrey S. Morris Texas A&M University 2000
Hua Liang Texas A&M University 2001
Inyoung Kim Texas A&M University 2002
Chan Hee Jo Texas A&M University 2003
Tanya Apanasovich Texas A&M University 2004
Gosia Leyk Texas A&M University 2004
Christie Spinka Texas A&M University 2004
Veera Baladandayuthapani Texas A&M University 2005
Yehua Li Texas A&M University 2006
Iryna Lobach Texas A&M University 2006
Bo Li Texas A&M University 2006

xvii
xviii PhD Students of Raymond J. Carroll

Name Institution Year

Lian Liu Texas A&M University 2007

Arnab Maity Texas A&M University 2008
Seokho Lee Texas A&M University 2009
Andrew Redd Texas A&M University 2010
Jiawei Wei Texas A&M University 2010
Saijuan Zhang Texas A&M University 2010
Trijya Singh Texas A&M University 2011
Xiaolei Xun Texas A&M University 2012
Contributors

John P. Buonaccorsi
University of Massachusetts, Amherst, MA, USA

George Casella
University of Florida, Gainesville, FL, USA
(deceased)

Aurore Delaigle
University of Melbourne, Melbourne, Australia

Laurence (Larry) Freedman

Gertner Institute for Epidemiology, Tel Hashomer, Israel

Mitchell H. Gail
National Cancer Institute, Bethesda, MD, USA

Roger Koenker
University of Illinois, Urbana-Champaign, IL, USA

Yehua Li
Iowa State University, Ames, IA, USA

Hua Liang
The George Washington University, Washington, DC, USA

Dale L. Preston
Hirosoft International, Eureka, CA, USA

xix