1. With an educational background in mathematics and statistics at City College of New York and University of North Carolina, when and how did you first become aware of statistics as a discipline?

In my younger days, we played many card games and bet on sporting events. I saw the connection between probability and my usual card playing and betting. The probability course was one of the few college courses that connected with my real life.

2. What was it that inspired you to pursue a career in biostatistics?

I had an appointment at the University of Wisconsin in a Math Centre. There were no restrictions on my activities. I had a “very long term fellowship” with no duties. A medical investigator presented me with data from a clinical trial. Using that motivation, my friend Georg Weiss and I developed the theory of semi-Markov processes. Unfortunately we delayed publication of our developments for many years and did not get the credit we would have received with an earlier publication. During my time in Wisconsin I had some surgery, which did not go well. The sutures broke and luckily I woke up and called for assistance. I thought that I would like to learn more about applications in medicine and biology and this “scare” prompted me to leave University of Wisconsin to learn more about biostatistics. I thought life could be too short.

3. It has been said that ‘before Marvin Zelen came along, biostatistics had something of a Rodney Dangerfield problem: it rarely got the respect it deserved’ (Peter Wehrwein) and that you have ‘transformed clinical research’ by making it an organised and statistically sophisticated area of medical research. What do you personally feel have been your main contributions to the world of biostatistics and what achievements are you most proud of?

Of course this is a great exaggeration! Perhaps my contribution was to recognize the need for careful organization of trials and the need for stressing data quality. I invented the word “data manager”. Also I think my emphasis on paying careful attention to the scientific design of studies was important. I was the Principal Investigator of a grant to bring community hospital cancer patients into cancer clinical trials. Up to that time patients mainly came from large centers and university hospitals. Although the National Institutes of Health gave me the grant, a review committee had originally turned it down as they felt that community hospitals could not deal with entering patients on clinical trials. This new program created a major change in cancer trial design as the community hospital patients tended to be newly diagnosed patients. The university hospitals and large treatment centres tended to have patients who had failed on initial therapy. Another thing I did was introduce e-mail to the clinical trial investigators. In 1977 there was no such thing as e-mail. In my group we wrote software to use within e-mail. There was no such thing as an internet. When I first submitted a grant for writing an e-mail package, one evaluation called it “stupid” and would never work with clinical investigators. After two rejections we went ahead anyway (using other funds) and the introduction of the e-mail greatly added to our ability to introduce better studies.

When I first submitted a grant for writing an email package, one evaluation called it “stupid!” and would never work with clinical investigators. After two rejections we went ahead anyway (using other funds) and the introduction of the email greatly added to our ability to introduce better studies.

4. You have worked at the National Bureau of Standard and Technology’s Statistical Engineering Laboratory and the National Cancer Institute’s Applied Mathematics and Statistics section. What was it that led you to form the Statistical Laboratory at the University of Buffalo and how did your previous roles bear impact on this move?

It was one way to deal with the University bureaucracy at Buffalo. Also some members of my academic department did not think this was suitable activity for an academic department. At one time our Lab was the cancer clinical trial centre of the world. We were coordinating trials for: the Veterans Administration Lung Cancer Study Group, The Eastern Cooperative Oncology Group, The Radiation Therapy Oncology Group, and The Colon Cancer Oncology Group. We also help organize the Statistical Centre of the European Organization of Cancer.

5. In the 1980s, you became very well-known by helping to find that there was a link between a number of childhood leukaemia cases in recent years and the contaminated water supply of a suburb called Woburn in Boston. How exactly did you interpret these findings in your data? What kind of methods did you use to find this link?

With my colleague, the late Steve Lagakos we showed that the incidence of child leukaemia was proportional to the amount of contaminated well water going to the households. The control group were children without leukaemia. We obtained data about the water distribution in the town by chronological year and month and thus were able to make such an analysis. We did not rely on the higher incidence to draw our conclusions. We also, in cooperation with the residents, did a telephone survey to determine birth defects. We found some classes of birth defects were related to the amount of water coming from the contaminated wells. The analyses were unusual as most residents had a water supply which was a mixture of water coming from the 8 wells supplying the town. Some years the contaminated wells were turned off resulting in fewer birth defects for new-borns. It was a very political time. In the U.S., the super fund legislation was being renewed. Essentially it was a tax on the U.S. chemical industry. The Administration felt that it would make the US chemical industry less competitive. The renewal enlarged the legislation to provide victim compensation. Opponents claimed no one had been shown to be seriously “hurt”. Then our paper came along.

On the Woburn study –Opponents claimed no one had been shown to be seriously “hurt”. Then our paper came along.

6. You continue to teach at Harvard University as Lemuel Shattuck Research Professor of Statistical Sciences. 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?

I voluntarily changed my appointment over five years ago from a tenured professor to a research professor in order to step aside for a younger person. At my age, at that time, I did not think it was proper to still occupy a tenured position. I have not taught a class since then. The teaching of biostatistics is rapidly changing as internet courses are being offered reaching thousands of students. However it is not clear whether such courses are satisfactory for students pursuing a research career. I think the interaction between instructor and student is very necessary for students pursuing a research career.

7. Your current research focuses on the creation of stochastic models for the early detection of disease, and clinical trials, randomization and inference. What are your main objectives and what do you hope to achieve through your research?

I think the early detection of disease process is not well understood. Nevertheless it offers possibilities for saving lives with current treatments. I have tried to clarify some of these issues.

I believe that the current statistical methods for analysing clinical trials are not correct. The reason being we do not have a random sample of patients entering trials. All of our stat techniques require a random sample. As a result, in randomized trials the only uncertainty is the randomization process. It is possible to make the inference only depend on the randomization process resulting in greater power. However the inference is restricted to the better treatment for patients entering the trial. Not the better treatment for disease.

8. You have authored over 160 publications. Is there a particular article that you are most proud of?

I am very proud of several papers – among which is the paper with Weiss on Semi—Markov Processes, two papers on the early detection of disease , one with Manning Feinleib (1969) and the other with Sandra J. Lee (2008), a paper on randomization in clinical trials which laid out several of the randomization algorithms in current use; the Woburn paper with Steve Lagakos; a chapter with Norman Severo on numerical methods with respect to Probability Functions (it appeared in the NBS Handbook of Math Functions and Tables which is the largest distributed book in the Math Sciences) and a paper on evaluating the participation of community hospitals in cancer clinical trials (New England Journal of Medicine).

I believe that the current statistical methods for analysing clinical trials are not correct.

9. In relation to the above question, 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?

Yes!!! NO doubt about that.

10. Do you have any advice for students considering a university degree in statistics?

My advice is to be very skilled in modern computing. It is clear to me that computing will supplant mathematics as the key underlying method relating to advances in statistics/biostatistics.

11. 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?

Motivation often comes from new consulting problems as well as when one is teaching requiring re-examining assumptions and formulations. I have always tried to introduce new ideas in my teaching. Very often, a take home exam contains original research problems with hints on solutions.

12. What has been the most exciting development that you have worked on in biostatistics during your career?

The Woburn study was very exciting. I also have to admit, out of vanity, that the first time I was invited to make a presentation at a scientific meeting – I was very thrilled.

13. 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 most important developments have been the growth of computing and how it has affected new developments in biostatistics and the ability to analyze large data sets.

I think the future will be very much related to “Big Data”; i.e. Large amounts of observational data, collected for other purposes, which can be used to learn about phenomena.

14. What do you see as the greatest challenges facing the profession of biostatistics in the coming years?

The greatest challenge is to attract more talented students to the profession.

15. Are there people or events that have been influential in your career?

Yes—I was very fortunate to obtain a position at the National Bureau of Standards when I was quite young. In fact I was the only one in the Math Division without a doctorate. I also think I was the youngest. My stay there was very influential. Among my mentors were Churchill Eisenhart and Jack Youden. In my first day on the job, the head of the Division called me in (he was a famous American mathematician) and asked me if I wanted to be happy on the job. “Yes Sir!” I said. I was 25 years of age and the only one without a doctorate. He said,” Do good research and lots of it”. Also I looked about 15 years of age.