George Alfred Barnard

From Wikipedia, the free encyclopedia

George Alfred Barnard (September 23, 1915 - August 9, 2002) was a British statistician known particularly for his work on the foundations of statistics and on quality control.

Contents

[edit] Biography

George Barnard was born in Walthamstow, London. His father was a cabinet maker and his mother had been a domestic servant. George attended the local grammar school, the Monoux School, and from there he won a scholarship to St John's College, Cambridge to read mathematics. In 1937 he went on to Princeton University to do graduate work on mathematical logic with Alonzo Church. Shortly before Alan Turing had gone from Cambridge to study with Church.

Barnard was on holiday in Britain when the Second World War started and he never went back to Princeton to finish his PhD. The war made Barnard into a statistician as it did for many mathematicians of his generation. In 1940 he joined an engineering firm, Plessey, as a mathematical consultant. In 1942 he moved to the Ministry of Supply to apply quality control and sampling methods to the products for which they were responsible. It was there that Barnard began doing statistics. The group he was put in charge of included Peter Armitage, Dennis Lindley and Robin Plackett. Lindley recalls that they were like students working for a doctorate with Barnard as supervisor. Abraham Wald was in a similar group in the United States. Both groups developed sequential methods of sampling.

At the end of the war, Barnard went to Imperial College London, as a lecturer, becoming a reader in 1948 and professor of mathematical statistics in 1954. In 1966 he moved to the newly created University of Essex, from which he retired in 1975. Barnard, however, kept on doing statistics until he died aged 86. Until 1981 he spent much of each year at the University of Waterloo, Canada, and after that he continued writing papers and corresponding with colleagues all over the world.

Barnard's best known contribution is probably his 1962 paper on likelihood inference but the paper he thought his best was the 1949 paper in which he first espoused the likelihood principle. He had originally described the principle in the context of optional stopping. A statement by Savage brings out how surprising the principle first seemed (Foundations of Statistical Inference, 1961, p. 75)

I learned the stopping rule principle from Professor Barnard in … 1952. Frankly, I then thought it a scandal that anyone in the profession could advance an idea so patently wrong, even as today I can scarcely believe that some people resist an idea so patently right.

Barnard’s first publication was “A New Test for 2X2 Tables” (1945). The old test was Fisher's and Fisher was not pleased! However he convinced Barnard that the new improved test was inferior to the old and the two became friends. Barnard had actually met Fisher in 1933 just before he left school. Barnard liked telling the story of their meeting. Barnard had done a survey of the political attitudes of sixth-formers and how they had formed them. He ended up going to Fisher for advice on analyzing the results. Fisher showed him Statistical Methods for Research Workers and said, “if you read this book, you’ll find a lot of statements in it that are made without proof. You’re a mathematician, you should be able to prove them for yourself. And if you’ve done that, you’ll know statistics.” Barnard found the last piece of the puzzle nearly twenty years later—just as he was becoming vice-President of the Royal Statistical Society to Fisher’s President. The story could be a metaphor for a side of Barnard's work. He kept returning to Fisher's work, trying to clarify his recondite ideas on likelihood, fiducial probability and pivotal inference. Barnard came under Fisher's spell just as his star was fading for the younger generation of statisticians and he remained Barnard’s great hero amongst statisticians.

In an interview Barnard recalled, “my main interest above everything was politics from about 1933 until 1956. Well, that’s not true—until the end of the war it would be fair to say.” At school he proposed the motion to the school debating society that “Socialism is preferable to Capitalism." He joined the Communist Party in 1933 and he took part in anti-fascist marches in the east end of London. At Plessey he was chairman of the shop stewards.

Barnard served terms as president of three societies, the Royal Statistical Society in 1971-2, the Operational Research Society in 1962-4 and the Institute of Mathematics and its Applications in 1970-1. He was awarded the Guy Medal in Gold by the Royal Statistical Society in 1975.

George Barnard was very open-minded and very well liked. Lindley wrote in The Statistician, “We have lost a great statistician and a delightful human being. “

[edit] Publications

There is a bibliography (containing 109 articles) up to 1989 in

  • Seymour Geisser et al. (eds) Bayesian and Likelihood Methods in Statistics and Econometrics : Essays in Honor of George A. Barnard, North-Holland 1990.

This contains a review of Barnard's work by Lindley. The volume was one of a series honouring Bayesian heroes. Barnard was not a Bayesian but he was a "great guy"!

In 1990 he made a book out of manuscripts left by his friend Egon Pearson

  • E. S. Pearson (1990) ‘Student’, A Statistical Biography of William Sealy Gosset, Edited and Augmented by R. L. Plackett with the Assistance of G. A. Barnard, Oxford: University Press.

After 1990 Barnard published little, although he kept up his letter writing. In 1996 however he produced a review of Barndorff-Nielsen and Cox

  • Review of Inference and Asymptotics. by O. E. Barndorff-Nielsen; D. R. Cox Journal of the Royal Statistical Society. Series A, 159, (1996), 178-179.

After observing that, “A great virtue of the book is that it raises perhaps as many questions as it answers,” Barnard went on to give his answer one of those questions.

[edit] References

[edit] See also

[edit] External links