|
Dr.
Stephen E. Fienberg is the Maurice Falk University Professor of
Statistics and Social Science in the Department
of Statistics, the Machine
Learning Department and Cylab
at Carnegie Mellon University in Pittsburgh, Pennsylvania.
He
received the B.Sc. (Mathematics and Statistics) degree from University
of Toronto in 1964, the A.M. (Statistics) degree from Harvard University
in 1965 and the Ph.D. (Statistics) degree from Harvard University in
1968.
Among
many honors, Dr. Fienberg is an Elected
Member of The National Academy of Science, a Thorsten Sellin Fellow of
the American Academy of Political and Social Science (2004),
and Elected fellow of the Royal Society of Canada, the American
Association for the Advancement of Science, the American
Statistical Association, and the Institute of Mathematical
Statistics.
He was
the President (1998-1999) of the Institute of Mathematical Statistics.
He is an Editor of the Annals of Applied Statistics for Social Science,
Government and Economics (2006-). He is the author and co-author of
numerous research articles in leading international journal, books, and
special volumes. More details can be obtained from the site www.stat.cmu.edu.
Dr.
Fienberg's current research interests include:
- Analysis of categorical data; Bayesian approaches to
confidentiality and data disclosure; causation; foundations of
statistical inference; history of statistics; sample surveys and
randomized experiments; statistics and the law; inference for
multiple-media data.
- His principal research interests lie in the development of
statistical methodology, especially for problems involving
categorical variables. Initially, he worked on the general
statistical theory of loglinear models for categorical data, and he
applied the theory to various problems that could be represented in
the form of multidimensional contingency tables. More recently, he
has studied approaches appropriate for disclosure limitation in
multidimensional tables and their relationship with results on
bounds for table entries given a set of marginals (for selected
publications on this topic see Disclosure
Limitation Papers, as well as the webpage for the NISS
Digital Government Project on this topic),
estimating the size of populations (especially in the context of
census taking), and Bayesian approaches to the analysis of
contingency tables. His research on disclosure limitation for
categorical data, and on privacy and confidentiality more generally,
has led to the creation of a new online journal, The
Journal of Privacy and Confidentiality, which has just begun to
accept submissions.
- For some interesting historical material on the model for
quasi-symmetry and the work of Henri Caussinus, see Project
QS,and the special issue of Annales
de la Faculté des Sciences de l'Université de Toulouse Mathématique
in honor of Caussinus dated 2002.
- For several years now, he has also worked on the development of
statistical methods for large-scale sample surveys such as those
carried out by the federal government. This work (much of which has
been in collaboration with Judith Tanur) has included the study of
nonsampling errors, the use of surveys to adjust census results for
differential undercount, cognitive aspects of the design of survey
questionnaires, statistical analysis of data from longitudinal
surveys, and formal parallels in the design and analysis of sample
surveys and randomized experiments. His recent book with Margo
Anderson, Who Counts? (which has now appeared in a revised
paper back edition), chronicles the story of the the 1990
decennial census and efforts to use sample to adjust census
results for differential undercount. His work on
confidentiality and disclosure limitation ties both to surveys and
censuses and also to categorical data analysis (again see the
webpage for the NISS
Digital Government Project on this topic as well as some
of the selected papers below), and also addresses public concerns
about privacy. For a July 2001 news story on the topic of
privacy in the Pittsburgh
Post-Gazette, click here.
- In the analysis of data from longitudinal studies of
disability, such as the National Long Term Care Survey, a number of
authors have used novel statistical methodology based on what has
come to be known as the Grade of Membership (GoM) model. Working with students and colleagues, I have been exploring the GoM
model, its estimation, and comparisons between it and other
categorical data models. We have also begun to look at
confidentiality issues arising in the context of the NLTCS. Some of our work is available on a separate webpage: NLTCS,
the GoM Model, and Confidentiality.
- He has also been active in the application of statistical methods
to legal problems and in assessing the appropriateness of
statistical testimony in actual legal cases, and he has linked his interests in Bayesian decisionmaking to the issues of legal
decisionmaking. For information on the NAS Sackler Symposium
on Forensic Science, held November 16-18, 2005.
|
