Peter Bickel is an American statistician and Emeritus Professor of Statistics in Berkeley. He is past President of the Bernoulli Society and of the Institute of Mathematical Statistics.
He studied physics for two years at the California Institute of Technology and then graduated with a BA and MA in mathematics and Ph. D in statistics in 1963 from University of California, Berkeley, where his advisor was Erich Leo Lehmann.
As a scientist and an educator, he has supervised and mentored many students and young researchers, many of whom later become leading figures in the fields of statistics and economics, such as C.F. Jeff Wu, Jianqing Fan,Y.Ritov , Mark van der Laan and Donald Andrews.
His current research interests involve developing network theory and applying statistics to biology. Professor Bickel talked to Statistics Views during JSM 2013 in Montreal, Canada.
1. You are currently Emeritus Professor of Statistics at the University of California at Berkeley. Over the years, how has your teaching and research motivated and influenced each other? Do you continue to get research ideas from statistics and incorporate them into your teaching?
One of the conditions I gave when I retired was that I would not have to teach undergraduates. Whilst I wanted to I just couldn’t do it very well. I can be sloppy and make mistakes and undergraduates are not very tolerant of that! I enjoy PhD supervision and supervise many students. My research ideas emerge a lot through interaction with these students. I count on presentations of my work to attract them.
2. You are a former President of both the Bernoulli Society and the Institute of Mathematical Statistics. What are your memories when you look back on your Presidencies and what do you feel were your main achievements?
I was President of the IMS in 1980 and I remember we had almost an all-night crisis meeting because too many papers had been accepted by our journals and we were not able to afford the revenue that would have been required that these papers be published promptly . With such an enormous backlog, it would be years before many of these papers could be published, so the solution, which I didn’t entirely agree with, was to publish the papers in a very small typeface! Luckily the profession was young and had good eye sight, so we didn’t get many complaints.
With the Bernoulli Society, I wanted to be able to help with a situation in the Middle East . The Israelis were keen on having a World Bernoulli Congress in Israel but this was just before the time of the Yom Kippur War, so tensions were very high. Although I have never believed in boycotts, the majority of officers felt it would not have been safe. This was something that had to be let go, which was a shame.
3. How do you think the Bernoulli Society and the IMS have evolved over time and adapted to the changing needs of the statistical community?
The Bernoulli Society has always been more mathematical. For the record, I also belong to the Royal Statistical Society which reflects, if you like, the more ‘Fisherian’ approach in applying theory and the ASA. I think both societies (IMS and Bernoulli) have tried to become more international and the IMS, for instance, now has a regional subsection in China and the Far East. Membership has grown for both. One feature they have both maintained is a closer link to mathematics through probability than the other societies. Both started as purely mathematical but have since grown in terms of worldwide coverage and attracting the interest of computer scientists and other domains of science.
4. What has been the most exciting development that you have worked on in statistics during your career?
Many statisticians proceed by finding a theme and sticking to it. That is not my style and I tend to go around and make, I think, good contributions. In many areas my goal is to do work which is both potentially influential and also mathematically fun to carry out. I tend to get distracted. I do think my work in semiparametric statistics has been the most exciting so far. I hope more recent research that has a bearing on biology which I find very exciting will turn out to be influential as well.
The process of finding such representations and making valid inference about them is, I think, the most exciting area of theoretical statistics these days.
5. What do you think the most important developments in the field have been? What do you think will be the most exciting and productive areas of research in statistics in the coming years?
I do think that network data is a key element in understanding complex systems. It is not well understood and there are all sort of surprising phenomena. More generally exploring what can happen in high dimensional data and seeing to what extent you can draw conclusions are also both important. One issue that I mentioned in the Fisher lecture is that there is a very puzzling thing in that in statistical theory, it is quite apparent that if you assume nothing, e.g. want to fit a curve but all you know is that it has a certain amount of smoothness, you quickly come to the conclusion that if ,in fact, God is malignant, you’re in big trouble. You have to have numbers of observations that start to resemble the number of atoms in the universe in order to get hold of the signal in this data.Yet we are able to do all sorts of things.
There is a general sense that, in fact, God is not malignant but subtle, as Einstein said, and that most data somewhere, somehow have a lower dimensional representation. But the difficulty is in finding it. Once you’ve found it, you are in some sense back in Fisher’s world – the only problem then is that the process of finding it involves dealing with the data quite intimately, so that by the time you’re finished, it’s not so clear that inferential statements are valid. There is a great temptation to say that you knew it all along and then you compute on that basis but that may be wishful thinking.
The process of finding such representations and making valid inference about them is, I think, the most exciting area of theoretical statistics these days.
6. What do you see as the greatest challenges facing the profession of statistics in the coming years?
I think it has always been the same challenge in some sense. Statistics on the one hand pervades everything and because it does, other subject areas claim statistics as part of their arena. So I think maintaining the distinct nature of being a statistician has been the main challenge and remains so. What I’ve found is it’s better not to worry so much about rubrics but more about that when you’re involved in a project, you really have to get to the heart of it and at least be able to understand what the main questions are. I think that there has unfortunately been a period where to some extent the statistician was viewed only as a technical consultant. This is neither intellectually stimulating nor a proper representation of what we’re capable of.
7. Are there people and events that have been influential in your career?
Of course, yes. My thesis advisor, Eric Lehmann; and his collaborator Joe Hodges whose class I took. Hodges was a very talented applied statistician, although he really prized doing hard mathematics. But when he did work on statistics, it was clear that he really understood what was going on!
Lehmann and Hodges drew me from mathematics to statistics and fostered my career at Berkeley where I have stayed ever since.
Like many high school students of my generation, I was influenced by the books of Georges Gamow, a physicist who originated the Big Bang Theory, and played an important conceptual role in Crick’s discovery of the DNA-amino acid code. His books did turn me towards physics. Although I didn’t end up as a physicist, I believe that the spirit of statistics is close to that of physics. So I landed close to my initial attractor after all.
I had an uncle who was a lawyer and wrote for a major newspaper in New York. He was a very interesting person whom I loved to visit. My father, who was a physician, died when I was two years old and although my mother remarried, my uncle continued to play a big role in my life.