Cambridge University Reporter


Report of Discussion

Tuesday, 7 March 2006. A Discussion was held in the Senate-House. Deputy Vice-Chancellor Dr Kate Pretty was presiding, with the Senior Proctor, the Junior Proctor, a Pro-Proctor, the Registrary, and five other persons present. The following Report was discussed:

Report of the General Board on the establishment of a Professorship of Statistics (p. 406).

Professor G. GRIMMETT:

Deputy Vice-Chancellor, I rise as Head of the Department of Pure Mathematics and Mathematical Statistics, and as a probabilist with the highest aspirations for statistics at Cambridge University. I welcome this Report. This is the cornerstone of a programme to invigorate statistics within the Mathematics Faculty. This programme will develop a culture in which core statistics can flourish, while looking outwards towards the statistical needs of applied science.

An earlier attempt to develop Mathematical Statistics in Cambridge (I speak here of Modern History, not Ancient) encountered difficulties through limited resources and commitment. The present programme has the full support of the School of the Physical Sciences, and will be aided by significant external grant income. Both Department and School are committed to its success, and to making available the necessary resources. The programme enjoys also the support of the many scientific communities in Cambridge where statistics is applied.

When reflecting on Cambridge Statistics, one is led inevitably to a consideration of Cambridge Mathematics. Our new buildings are the envy of the mathematical world. For Cambridge, they are an opportunity. The unity of the Centre for Mathematical Sciences confronts the divided tree of Cambridge Mathematics. George Batchelor is credited with the formation of the Department of Applied Mathematics and Theoretical Physics in 1959. Shortly afterwards, the more ancient Statistical Laboratory amalgamated with Pure Mathematics to form my Department under William Hodge. Much excellent science has been done since, but some areas, including statistics, have encountered serious problems despite the very high level of collaboration and co-operation between the two Departments. It is time to reconsider the distinctions drawn by our predecessors, and to discard them where they no longer serve us well.

Statistics will develop best within a unified Department of Mathematics, as an equal partner to Pure Mathematics, Applied Mathematics, and Theoretical Physics.

The first step for statistics is to establish this new Professorship, and I commend the Report to the University.

Professor Sir JOHN KINGMAN:

Madam Deputy Vice-Chancellor, I was very pleased to hear the remarks of Professor Grimmett this afternoon. My pleasure was not shared by many members of the Regent House: evidently, statistics is too boring even for Professor Evans. I am no statistician, but I have watched the development of statistics in Cambridge since I was a student and then a Lecturer in the Statistical Laboratory, later as an external Elector to various Professorships, and now as a member of the Council of the School of the Physical Sciences. In this last capacity, I strongly support the recommendations of the Report now before us, but in doing so I utter a warning.

There is a history, as Professor Grimmett has acknowledged. Cambridge had no Professor of Statistics until 1962, although the formidable Sir Ronald Fisher, the father of modern statistics, had occupied the Chair of Genetics from 1943 to 1957. There was some surprise in 1962 when the University, having created a Chair of Mathematical Statistics, elected to it a probability theorist rather than one of the then active statisticians (many of them Cambridge graduates). However, Professor Kendall was so distinguished that criticism was muted, and he later developed his own original approach to data analysis which could be seen as additional justification for the choice.

There was much greater concern in the statistical community when, on Kendall's retirement, his successor turned out to be another probability theorist, and then, when he in turn left Cambridge, his successor was also a probabilist. Meanwhile another Chair had been created in the Laboratory, in the mathematics of operational research, but this was never intended to cover statistics, and both holders of that Chair have worked in other areas. Cambridge has become a major centre of excellence in pure and applied probability, but, at the same time, it has given the outside world a clear signal that the University regarded statistics as a much lower priority.

There was an attempt to remedy this in the early 1990s with the creation of a new Chair of Statistical Science. The Report proposing that post reads very much like the Report which is before the Regent House today. The first holder of this third Chair in the Laboratory was indeed a mainstream statistician, but he found the Cambridge environment so uncongenial that after two years he returned to his previous post in the USA. The Chair was vacant for some years, and was then filled by a probability theorist, one who applied the theory to problems of finance.

Thus, of some 111 professorial man years of the Statistical Laboratory (including three ad hominem Professors) only three were occupied by mainstream statisticians. Three out of 111 is not an impressive success rate. Cambridge therefore has a credibility problem in filling the new Professorship now proposed. Why should any good statistician apply, when he or she knows that Cambridge has such a record of advertising for statisticians but then appointing probabilists?

It is therefore crucial that the University commit itself to recruiting a real statistician to this new position. Good intentions and fair words are not enough. I speak from a certain first-hand experience of professorial recruitment when I say that it will require a sustained effort at all levels, up to and including the Vice-Chancellor, if the right appointment is to be achieved.