Search

The popularity of the EM algorithm owes much to the 1977 paper by Dempster, Laird and Rubin. That paper gave the algorithm its name, identified the general form and some key properties of the algorithm and established its broad applicability in scientific research. This review gives a nontechnical introduction to the algorithm for a general scientific audience, and presents a few examples characteristic of its application., Published in at http://dx.doi.org/10.1214/08-STS270 the Statistical Science (http://www.imstat.org/sts/) by the Institute of Mathematical Statistics (http://www.imstat.org)

In this paper, we review analytic methods for regression mod- els for longitudinal categorical responses. We focus on both likelihood- based approaches and non-likelihood approaches to analysing repeated binary responses. In both approaches, interest is focussed primarily on the regression parameters for the marginal expectations of the binary responses. The association or time dependence between the responses is largely regarded as a nuisance characteristic of the data. We consider these approaches for both the complete and incomplete data cases. We describe the generalized estimating equations (GEE) approach, a non-likelihood approach, and some proposed extensions of it. We also discuss likelihood-based approaches that are based on a log-linear repre- sentation of the joint probabilities of the binary responses. We describe how a likelihood-based "mixed parameter" model yields likelihood equa- tions for the regression parameters that are of exactly the same form as the GEE. An outline of the desirable features and drawbacks of each approach is presented. In addition, we provide some comparisons in terms of asymptotic relative efficiency for the complete data case, and in terms of asymptotic bias for the incomplete data case. Finally, we make some recommendations concerning the application of these methods.

Florence Nightingale David was born on August 23, 1909 in Ivington, near Leominster, England. She received her degree in Mathematics in 1931 from Bedford College for Women. In 1933 she became Research Assistant to Karl Pearson at University College, and in 1935 she was appointed Assistant Lecturer in the Statistics Department, University College, London. She received her doctorate in Statistics from University College in 1938. During World War II, she served as Experimental Officer in the Ordnance Board for the Ministry of Supply, Senior Statistician for the Research and Experiments Department for the Ministry of Home Security, member of the Land Mines Committee of the Scientific Advisory Council and Scientific Advisor on Mines to the Military Experimental Establishment. She returned to the Statistics Department at University College in 1945 where she was appointed Lecturer, Reader and then Professor in 1962. Beginning in 1958, she made regular visits to the United States, principally as Visiting Professor and Research Statistician at the University of California at Berkeley with the Department of Statistics, and the Applied Climatology and Forestry Divisions. She was elected to the International Statistical Institute, Fellow of the American Statistical Association, Member of the University Senate at University College, Governor of Bedford College for Women, and served as Review Editor for Biometrika. In 1968 she became Professor and then Chair of the Department of Biostatistics, at the University of California in Riverside. In 1970 when the Department of Statistics was created, she became Professor and Chair of Statistics. She retired from Riverside in 1977 and moved to Berkeley where she continues to be active as Professor Emeritus and Research Associate in Biostatistics. F. N. David is the author of nine books (a tenth is in progress, on the measurement of natural populations), two monographs and over 100 papers in scientific journals. Many of these are actively referred to today. Combinatorial Chance (with D. E. Barton) contains fascinating combinatorial probability theory (much like Feller's Volume 1) impossible to find elsewhere. Her Tables of Symmetric Functions contains a 50-page introduction that is still the standard reference. Her Probability Theory for Statistical Methods contains the only available treatment of "Lexis Theory," a forerunner of contingency table analysis. Her book, Games, Gods and Gambling, is widely recommended as an entertaining and authoritative account of the history of probability. Citations for these and her other five books are at the end of this article. Some of her 100 published papers 1932-1976 involve joint work with co-authors, including notables K. Pearson, J. Neyman, N. L. Johnson, M. G. Kendall, D. E. Barton, C. L. Mallows and E. Fix. To become further acquainted with her work, we can warmly recommend the following semester-long seminar: go to the Index to Statistics and Probability (Ross and Tukey, eds.), look up F. N. David and read through a selection of her papers. This interview took place in the Jerzy Neyman Conference Room at University of California in Berkeley in July 1988.