Upload
overlord303
View
20
Download
0
Embed Size (px)
DESCRIPTION
Short article on the maximum likelihood approach in statistical analysis
Citation preview
arX
iv:0
804.
2996
v1 [
stat
.ME
] 1
8 A
pr 2
008
Statistical Science
2007, Vol. 22, No. 4, 598620DOI: 10.1214/07-STS249c Institute of Mathematical Statistics, 2007
The Epic Story of Maximum LikelihoodStephen M. Stigler
Abstract. At a superficial level, the idea of maximum likelihood mustbe prehistoric: early hunters and gatherers may not have used the wordsmethod of maximum likelihood to describe their choice of where andhow to hunt and gather, but it is hard to believe they would have beensurprised if their method had been described in those terms. It seemsa simple, even unassailable idea: Who would rise to argue in favor ofa method of minimum likelihood, or even mediocre likelihood? Andyet the mathematical history of the topic shows this simple idea isreally anything but simple. Joseph Louis Lagrange, Daniel Bernoulli,Leonard Euler, Pierre Simon Laplace and Carl Friedrich Gauss areonly some of those who explored the topic, not always in ways wewould sanction today. In this article, that history is reviewed fromback well before Fisher to the time of Lucien Le Cams dissertation.In the process Fishers unpublished 1930 characterization of conditionsfor the consistency and efficiency of maximum likelihood estimates ispresented, and the mathematical basis of his three proofs discussed.In particular, Fishers derivation of the information inequality is seento be derived from his work on the analysis of variance, and his laterapproach via estimating functions was derived from Eulers Relationfor homogeneous functions. The reaction to Fishers work is reviewed,and some lessons drawn.
Key words and phrases: R. A. Fisher, Karl Pearson, Jerzy Neyman,Harold Hotelling, Abraham Wald, maximum likelihood, sufficiency, ef-ficiency, superefficiency, history of statistics.
1. INTRODUCTION
In the 1860s a small group of young English intel-lectuals formed what they called the X Club. Thename was taken as the mathematical symbol for theunknown, and the plan was to meet for dinner oncea month and let the conversation take them wherechance would have it. The group included the Dar-winian biologist Thomas Henry Huxley and the so-
Stephen M. Stigler is the Ernest DeWitt Burton
Distinguished Service Professor, Department of
Statistics, University of Chicago, Chicago, Illinois
60637, USA e-mail: [email protected].
This is an electronic reprint of the original articlepublished by the Institute of Mathematical Statistics inStatistical Science, 2007, Vol. 22, No. 4, 598620. Thisreprint differs from the original in pagination andtypographic detail.
cial philosopher-scientist Herbert Spencer. One eve-ning about 1870 they met for dinner at the AthenaeumClub in London, and that evening included one ex-change that so struck those present that it was re-peated on several occasions. Francis Galton was notpresent at the dinner, but he heard separate ac-counts from three men who were, and he recordedit in his own memoirs. As Galton reported it, dur-ing a pause in the conversation Herbert Spencersaid, You would little think it, but I once wrote atragedy. Huxley answered promptly, I know thecatastrophe. Spencer declared it was impossible,for he had never spoken about it before then. Huxleyinsisted. Spencer asked what it was. Huxley replied,A beautiful theory, killed by a nasty, ugly littlefact (Galton, 1908, page 258).Huxleys description of a scientific tragedy is sin-
gularly appropriate for one telling of the history
1
http://arXiv.org/abs/0804.2996v1http://www.imstat.org/sts/http://dx.doi.org/10.1214/07-STS249http://www.imstat.orgmailto:[email protected]://www.imstat.orghttp://www.imstat.org/sts/http://dx.doi.org/10.1214/07-STS249
2 S. M. STIGLER
Joe Hodgess Nasty, Ugly Little Fact (1951)
Tn = Xn if |Xn| 1
n1/4
= Xn if |Xn|