Nicholas P. Jewell is Professor of Biostatistics and Statistics at the University of California, Berkeley. He was educated at the University of Edinburgh where he received a first class Honours degree in (Applied) Mathematics in 1973 and a PhD in Mathematics in 1976. Immediately following his graduate program he was appointed to a Harkness Fellowship from 1976-1978 which he held at the University of California, Berkeley and at Stanford University. From 1979-1981 he was an Assistant Professor of Statistics at Princeton University. He has also held academic appointments at the University of Edinburgh, University of Oxford, and at the Kyoto University. In 2007, he was a Fellow at the Rockefeller Foundation Bellagio Study Center in Italy, He is a Fellow of the American Statistical Association, the Institute of Mathematical Statistics, and the American Association for the Advancement of Science (AAAS). He is the 2005 winner of the Snedecor Award from COPSS, and won the Distinguished Teaching Award from UC
Berkeley’s School of Public Health in 2004. In 2012, he was awarded the Marvin Zelen Leadership Award.
StatisticsViews.com talks to Professor Jewell about his work in biostatistics, his pioneering work during the tragic AIDs epidemic and how he used statistics to prepare and secure University of California, Berkeley built directly on top of the San Andreas fault line for the inevitable.
1. With an educational background in Applied Mathematics at Edinburgh (a first-class BA and PhD respectively), when and how did you first become aware of statistics as a discipline?
I studied at Edinburgh in the late 1960s and in the first year, we had ten lectures in statistics. We then had ten lectures in computing, which back then was just starting. I was exposed to statistics from the very first week I started in Edinburgh. David Finney, who was Chair of Statistics, taught those first ten lectures but I have to say, I totally hated statistics! No disrespect to David at all but I was very much interested in maths, and did not like the inelegance and uncertainty of the mathematical language in statistics. So I never touched it again after that, nor did I take another course.
There was a strong department in statistics at Edinburgh and I became friendly socially with a number of faculty members and I started to become more conscious of statistics through conversations. I befriended a Mathematics professor named Sandy Davie who is still there and he was a very unusual and brilliant mathematician. He was very committed to solving applied problems, especially in the medical area. His work sparked my interest as my postgraduate training drew to a close. So I decided I would have another go at learning statistics.
Statisticians need to understand the uncertainty about big data and translate it so that accurate decisions can be made.
2. What was it that inspired you to pursue a career in biostatistics?
It was hard to get jobs in mathematics at universities in Britain then and I am not sure if that is still the case now. I was 23 when I finished my PhD which was a huge advantage as I could follow my career and move around, without having to worry about a family at that time. I wasn’t sure how good a mathematician I was. I was probably too harsh on myself at the time but I was concerned about restricting my interests – being in a room and solving problems that only a few people around the world would care about. I decided to do something more practical. I had always been interested in medicine but did not wish to become a doctor however, so biostatistics seemed to be the natural combination of all I wanted in terms of interests, society, people, problem-solving that people cared about, and mathematical skills. It was then I moved to the US on a scholarship where I planned to teach myself biostatistics. After a PhD in maths, people expected me to work in more theoretical statistics but I wanted to leapfrog that and do something applied.
3. You continue to teach at the School of Public Health at University California Berkeley as Professor of Biostatistics and Statistics and have taught at many reputable institutions such as Stanford, Princeton, Oxford and Kyoto. As a university professor, what do you think the future of teaching biostatistics will be? What do you think will be the upcoming challenges in engaging students?
It’s a terrific age for teaching statistics, the new ‘sexy job’ as Hal Varian said, Chief Economist at Google, who I got to know when we both served in the adminstration at Berkelet. A Harvard Business Reviewrecentlyexplored data science and the huge explosion of interest in statistics to the extent that people now believe that nearly every company should have a data scientist, essentially a statistician in disguise. So it’s a very different world from when I started off, and statistics is now widely seen as an important area of science, which is fantastic. In terms of being a professor, I think there is a big upswing in interest in students applying for statistics and one of the main challenges is “what do we teach these kids given the nature of what people are looking for in terms of future jobs?” That is tricky as interest is being generated a lot by big data problems and the wide availability of data. Statisticians need to understand the uncertainty inherent in big data and translate that effectively so that accurate decisions can be made.
The demands on statisticians are now greater, and the old-fashioned way of mathematics and classical statistics training I was exposed to may no longer work. Teaching statistics has not really changed over the years and needs to evolve, especially at the high school level, and the impact of computing needs to trickle down into first courses as that’s where the exciting opportunities lie. We are not all the way there. High school education of statistics has changed sufficiently over the past decade or so, so when undergraduate students arrive at university, there are gaps in their knowledge and universities are trying to help with this transition, an exciting endeavour in itself. In summary, how do we train the next generation of statisticians and yet at the same time not lose sight of the deeper understandings provided by mathematics?
How do we train the next generation of statisticians and yet at the same time not lose sight of the deeper understandings provided by mathematics?
4. Your current research focuses on statistical methods related to infectious diseases, biostatistical techniques in epidemiological data analysis, survival analysis and stochastic processes and genomics. What are your main objectives and what do you hope to achieve through your research?
The main thing I love about research is discovery! I love the underlying math which is where my talent lies, an area in which a lot of people have skills but not everyone, so you can make a living. What drew me to the field are discoveries that have made an impact in medicine and public health. I am very much motivated by actual scientific problems – what is the impact of this intervention, how can we estimate that, what data do we need, is there any data available that can help us to address these questions? And then the wonderful experience of taking what appears to be chaotic data, which all data is to some extent, and translating it into usable results. So much data can lead you in the wrong direction due to the biases involved. To be able to understand those and work to bring information together targeted at an important problem is what excites me about research. I still love sitting with a data set in front of me and a scientific question in mind, and then trying to extract information from this jumble of numbers.
I came to Berkeley from Princeton as a faculty member in 1981 which was at the very height of the HIV epidemic. We did not call it HIV at that point–it was early in our understanding but in terms of infections, but it was at its peak. There was an enormous amount of concern and we did not have an understanding on what the epidemic would grow into at all. We were faced with possibly the greatest infectious disease challenge ever. We were at the forefront, when we did not even know for certain that it was a virus–and people were trying to collect data on symptoms, etc.–and there was I was, a young statistician involved in this. Of course, it was an utterly tragic epidemic in San Francisco at the time and in many places now still, especially in third world countries. It is very bittersweet to say that it was exciting to be involved in the work to address the issues of the disease and how we could intervene. That’s what I always hope to do as a statistician–to address such serious issues and try to help any way I can.
5. You have authored many publications including the best-selling Statistics for Epidemiology. Is there a particular article or book that you are most proud of?
That’s so hard to say. I tend to get bored and what excites me is the first year in research on a new scientific problem. I’m struggling that year and I’m challenged but I enjoy it the most but after 3-4 years, I’m ready to move on.
It’s not necessarily a great characteristic to have. Many of my colleagues are absolutely dedicated to a particular problem for years and I’ve always felt guilty that I don’t have that persistence! It’s like I get good at tennis, and then I’ll try golf, then I get a bit good and get bored and move to handball.
I’ve always been proud of the articles I first wrote and ones where I was able to speak to renowned experts in the area. I don’t have a particular one but the very first one in mathematics I published stands out as I had never published before.
Statistics for Epidemiology was interesting to write after being in university administration for ten years when I had kept doing research but at a much lower level. Once I stepped out of administration and went back to being a full-time faculty member, I wanted to re-engage with the research world that had moved on during the past ten years. I had to get into a rhythm for my research and I wrote the book to get re-engaged and it helped to get me going again. Other papers that stand out are ones where the research took me by surprise or involved more hard work than usual. There is work I’ve done with students that I’m very proud of to see what they have accomplished.
I’ve always liked the paper I wrote on HIV early on in trying to convey how difficult it was to pass on the infection via sexual contact. There was a lot of misconception that all sexual contacts were equal with regards to HIV transmission. An Epidemiology student at Berkeley (Nancy Padian) had collected some new clear data on partnerships, where one partner had HIV but the other had not been infected. A student of mine (Steve Shioboski) and I were looking at HIV data on the partner and I was very interested about the statistical methods that could be brought to bear on the limited information we had—the results were different from pre-conceived notions.
We were faced with possibly the greatest infectious disease challenge ever. We were at the forefront, when we did not even know for certain that it was a virus–and people were trying to collect data on symptoms, etc.–and there was I was, a young statistician involved in this.
6. In 2006, you won the Director’s Award from the Federal Emergency Management Agency for “extraordinary leadership and vision in implementing strategies that enhance the disaster resistance of the University of California, Berkeley, and universities throughout America”. How did statistics play a part in these strategies?
The continuation of work at a time of a natural disaster is vital to the recovery of a community. In California, we are not prone to hurricanes but we are to earthquakes. Here at Berkeley, if an earthquake were to occur, and exams are about to take place, what do we do–close for six months? Some colleges in New Orleans never recovered after Hurricane Katrina and closed entirely. There is a very high probability of a major earthquake occurring on the fault line that runs right through the campus here at Berkeley. My office is literally within quarter of a mile within an area where large earthquakes have occurred in the past and there is a one in three chance of another major one within the next thirty years. Quake prediction is, of course, very difficult but you know it’s going to happen.
When I became involved in the campus administration, I never had any idea I’d be working on buildings and structures, but I happened to be given that charge. We had a long history of trying to retrofit buildings here and there. A past earthquake had not damaged the campus much, and trying to upgrade buildings was a long slow process. It’s hard to raise money for something for which there is no warning and you just have to be prepared.
At the time, there was a major earthquake in Kobe, Japan which caused much more damage to certain structures than had been anticipated. Possibly my statistical training gave me cause for concern at the randomness in dealing with this problem, so we hired in some engineering firms to do a survey in order to estimate the size of the problem rather than examining one building at a time.
This was in the early 1990s, five or so years after the last big earthquake in San Francisco. The results of the survey were shocking. Close to 30% of the buildings would pose a threat to a life in the event of a major earthquake, and the campus is a very densely populated area all year round. We had people working in buildings who would die if such an event occurred. That was the beginning. I wasn’t the most popular person on campus at that time because I brought attention to a problem that the administration could not ignore. Two chancellors I worked with were very much up to the task, and I suddenly got given the problem of solving it. I have no training in earthquake engineering so I sat down with my statistical background to lay out a strategy. We had to raise the money, but also how do we start to tackle this? Statistics came in by arguing about the average occupancy of a building per day.
I have no training in earthquake engineering so I sat down with my statistical background to lay out a strategy.
I wanted to be statistical about my decision and not just persuaded by deans, other forces, etc., so I tried to work on it from a perspective where human lives were the issue at stake in addition to business continuity. So we started building by building, retro-fitting and strengthening according to the strategy we developed. Although I subsequently left there is part of me that thinks that when the earthquake does eventually happen, whether I’m still here on earth or I’m looking down for above, we will know that our program has saved lives.
The other challenge was in approaching people in the buildings e.g. lab technicians who wanted to work in a safe building but didn’t want to be moved due to their experiments. There was a lot of initial resistance. I remember going to the Department of Chemistry having raised money to retrofit the building and I was very excited, but I was pretty much met by a chorus of boos. I was taken aback and I learned quickly that I had to sell it but make it look attractive at the same time. The retrofit and upgrade would not only provide a safer building but structures that would be better to do research in, and that was key to bringing faculty along with the plan that would bring disruption to their lives. It continues to be a challenge today and I have to say my one major failure in that is that the fault runs right underneath the stadium which can hold 70,000 people and you can see where part of the stadium wall was separated by a foot or two where the fault had shifted. It was clear to everyone that there was a problem that needed to be dealt with. I did the statistical calculations and the chances of someone being killed in the stadium were miniscule as it’s only used about six afternoons a year. No one worked there on a day-to-day basis at that time, and I held the line in making other campus buildings a higher priority. I said if the administration was worried, move the game to another stadium, advice which they did not like. After I left the administration, they rebuilt the stadium at enormous cost. I did not get across to them effectively that the chances of an earthquake hitting the stadium on a Saturday afternoon in November were negligible. Unfortunately a lot of people came back to me regarding the San Francisco 1989 earthquake that happened during the World Series just when the game was about to begin. No one was killed but it was in everyone’s minds as it had just happened. Statistics plays a central role in how we deal with crises and political events. That was my experience where I tried my best to apply statistical methods to policy decisions.
7. Do you think that statistics undergraduates and postgraduates starting out today are under more pressure to publish and to obtain grants than when you were a student yourself?
I don’t teach many undergraduates but mainly postgraduates. I think they have different pressures than I faced. I can remember having a conversation with a very senior mathematics professor at Edinburgh who advised me not to publish anything immediately after I got my first academic job, that it was more important to develop my courses, becoming a good teacher and letting my research mature like a good bottle of wine. I did not take him seriously and did publish before my PhD, but his advice today would be accepted as ludicrous. Postgraduate students today are expected to publish 2-3 papers whilst they are graduate students which was not required in my day.
The publish or perish movement was really picking up speed back then possibly due to a demand for academic positions and to hold on to these positions. Other institutions liked this model that students have to give evidence that you are good at doing research. I think it is way out of control now and that students are under too much pressure. I do not think anyone is smarter about research than they were when I was a student. It’s very hard to advise students about this as they feel they have to keep up with their peers.
A single person can only read a certain amount and with online publications, there may now be too much information out there with a danger that the really good research could be missed.
I would agree that there is more pressure, and there is less interest in academic positions, especially from young women, which really bothers me, as women often face academic pressures at a time when they are trying to start or raise a family. Pressures come and go with the economy and hopefully funding will pick up again, but it is very difficult at the moment.
8. Do you have any advice for students considering a university degree in statistics?
Go for it! It’s much more popular now with an excitement in the field and it relates to so many areas – biology, business and economics, political science, genetics – all these fields, and many more, recognise the importance of the discipline. I can’t imagine a better time for a young person to start a career in statistics as there are many exciting challenges.
Statisticians should also teach mathematics at universities where statistics isn’t that common but they can give different and exciting perspectives on the subject as compared to other mathematicians.
I can’t imagine a better time for a young person to start a career in statistics as there are many exciting challenges.
Statistics opens so many doors, including sports where metrics have become so popular.
9. Over the years, how has your teaching, consulting, and research motivated and influenced each other? Do you continue to get research ideas from biostatistics and incorporate your ideas into your teaching? Where do you get inspiration for your research projects and books?
Younger people do more energetic research and it’s one of the reasons I’d like to retire soon so someone can step into my shoes and have the opportunity to do their research. Consulting has informed my research, and pushing our field forward with original research also provides a backbone to my teaching–teaching and researching statistics are intrinsically linked.
10. What has been the most exciting development that you have worked on in biostatistics during your career?
By far the most important development was living through the HIV epidemic from the beginning as a statistician and not knowing what was happening and then having to figure everything out from data. It has been a devastatingly tragic epidemic for so many around the world. Just having the opportunity to work on this with my bare hands as a statistician and helping to piece together the puzzle of the epidemic–I look back on this period as an accomplishment. It was like being a relief worker in the aftermath of an earthquake, pulling away the rubble which felt awful yet at the same time, I was trying to help.
11. What do you think the most important recent developments in the field have been? What do you think will be the most exciting and productive areas of research in biostatistics during the next few years?
The field has really changed since I started so if I had to guess, the interest will be in effective methods of translating information and doing that with methods that are more complex than my generation were trained with handling.
The very objects of measurement now are changing and such tools need to be analysed to help with the advances that will be made. Computers are still continuing to change the face of statistics and now we have to figure out how not to drown under big data and to deal with it effectively.
…we have to try and put more resources into education and also to those who are learning statistics on the side so that when they go out into the world, they find that statistics is the career to pursue.
12. What do you see as the greatest challenges facing the profession of biostatistics in the coming years?
We’ve talked about needing to train the next generation to deal with recent advances and also adapt to different environments.
It is an exciting time to be a statistician. The opportunities are going to be enormous. But we are not producing enough statisticians to meet this new demand. Every company, every financial investment firm, every hedge fund, Amazon, Google, all want a statistician but there are not enough. Here in California, there are 2-3 programs on biostatistics and we have a handful of students. But biotech and pharmaceutical companies are now employing more than 100 statisticians. We can’t possibly meet this demand with current resources. The fraction of people needed in this work who are trained in statistics is falling while the number of jobs is growing.
More of these quantitative jobs are being taken up by computer scientists etc. who often lack a statistical background–we have to try and put more resources into education, also directed to those who are learning statistics on the side. When they go out into the world, they often find that statistics is a great career to pursue.
13. Are there people or events that have been influential in your career?
My advisor at Edinburgh was Alan Sinclair who always took the time to assist me and I owe him a great debt. At Stanford, I worked with a great group – Bill Bradley, Brad Efron, Persi Diaconis and Rupert Miller – I can’t imagine a better group to sit around and have lunch with. I didn’t know how famous they were but they were very kind and helpful.
At Princeton, Peter Bloomfield, John Tukey and the late Geoff Watson were very influential, especially Geoff. Jack Kalbfleisch at University of Michigan and I have quite parallel careers in statistics and administration and he has been enormously helpful.
Copyright: Photograph appears courtesy of Professor Jewell