David Dunson, winner of the 2010 COPSS Presidents’ Award, graciously agreed to respond to questions by Bhramar Mukherjee, COPSS secretary/treasurer.

Q1. What was your first reaction to winning the prestigious COPSS Presidents’ Award?

I had just looked online and found that it was a couple weeks past the date the committee was supposed to notify the winner, so I was a bit bummed and let Jim Berger know the sad news. Right after that, I heard from the committee chair, Harry Joe, that I had won, and Jim found it hilarious that he had to so quickly change his condolences to congratulations.

I’d like to say that I was calm and professional in hearing the news, but the reality was that I was overwhelmingly thrilled and ran around the house shouting, jumping up and down, and yelling like a complete lunatic. My wife, Amy, thought I had gone insane.

Q2. Which part of your job do you like the most?

There are many things I love about my job, but the best part is sitting in a coffee shop in the morning just getting started on a new and interesting problem with a pile of blank paper and no distractions. I really love all aspects of statistics research, but the best part is the initial problemsolving stage and the point at which you need to “dream up” fun new things to work on. It is important to maintain that free thinking and creative time, even with all the distractions that come with academia.

Q3. What advice would you give to young people who are entering the profession as PhD students and assistant professors at this time?

I would say that the most important things are to enjoy what you are doing, be creative, and take risks in choosing what to work on. Being a professor is really an awesome job and great lifestyle, with the freedom to set your own hours, work on what you like, and spend most of your time in a creative and dynamic environment surrounded by brilliant and interesting people. Instead of getting overly stressed out about all you need to accomplish to succeed and how difficult it is, just enjoy the process. If you love what you are doing, then productivity and success will come naturally.

In terms of choice of research topics, too much of statistics research is incremental and driven by a bandwagon effect. As a young researcher, don’t choose to work on topic A (e.g., large p, small n penalized regression) just because it seems that everyone else is and it is a hot area. Instead, read the literature and go to talks on a wide variety of subjects, having a skeptical eye and thoughts toward figuring out the limitations of what people are currently doing and how to do something better fully motivated by applications. Don’t get bogged down in the details of papers and seminars, but view it as a puzzle in which you win by figuring out a better big-picture way of doing things, instead of an incremental technical modification. Avoid writing papers that combine existing methods in obvious ways.

Q4. Who are your most significant mentors? How did/do they affect your career?

My tendency has been to follow my own path and figure out how to do things myself. This is not entirely a good character trait, and paying more attention to mentors along the way likely would have accelerated my early development as a researcher. I conducted my dissertation research on my own based on ideas I developed by reading papers on topics that seemed interesting and trying to think of better ways to do things.

My most significant mentor was clearly Clare Weinberg. She was my postdoc adviser and convinced me to stay at NIH in a tenure-track position. Clare is a great applied statistician and a brilliant and intense researcher. She has a talent for developing very simple statistical approaches for solving important applied problems and is naturally skeptical of modeling and complex methods. We had a wonderful dynamic in which I would attempt to come up with a flexible Bayesian hierarchical modeling approach, which she would then criticize. This taught me to be able to defend my ideas and to develop methods leading to real practical improvements in applications, lessons that have stuck with me and greatly increased the impact of my work.

Q5: Why were you drawn to Bayesian statistics?

I had no real exposure to Bayesian statistics during my graduate work, but I was bothered by the ad hoc nature of the frequentist solution I developed to an informative clustering problem in my dissertation. My initial interest in statistics came out of a desire to discover interesting new “biological truths” underlying high-dimensional, complex, and messy biological data, and such problems could not be addressed adequately in my view using frequentist methods. I did not believe in basing inferences on a point estimate, and I am skeptical of asymptotic justifications. In addition, we always have abundant prior information in science and it seemed ridiculous to ignore it. Hence, the Bayesian paradigm fits in perfectly with my philosophy of the “proper” way to do statistics, and it was natural for me to gravitate toward it early in my career.

My subsequent drift toward Bayesian nonparametrics is due to the thought that parametric Bayes hierarchical models are typically built on a house of cards of unverifiable modeling assumptions and it is wrong to assume that the true model is in a list of parametric models. Nonparametric Bayes is a young field, full of interesting problems, but already the methods have seen dramatic success in applications, particularly in machine learning.

Q6. Anything else you will like to share about our profession?

Just that I think I was extremely lucky to stumble into the field of statistics, which clearly has a daunting image problem. There are many fields that sound interesting to the general public, but what you are actually doing most of the time is not very stimulating intellectually. For example, work in many of the sciences involves very long hours of tedious busy work, and big discoveries are often based on luck.

The outside perspective is that what statisticians do must be extremely boring (memorizing baseball and health statistics, entering numbers, working through tedious calculations, etc.), but nothing could be further from the truth. In reality, we work with scientists in an amazing variety of fields and have a fundamental impact on how the studies are conducted and the results come out, while also having our own fascinating scientific discipline.

I had to stumble through engineering, geosciences, biology, and mathematics majors before discovering statistics, but we need to figure out—as a discipline—a better way to get the word out to top high-school and undergraduate students.

Q7: Finally, what are your hobbies/interests beyond statistics?

I think it is important to maintain active interests beyond academics to clear your head and maintain perspective. I personally love endurance sports, including trail running and open water swimming. I swim 2–3 miles with the Duke Aquatics team 5–6 days per week in the early morning and run about 30 miles a week. It was a blast swimming along the coast of Spain at the recent Valencia conference and running the trails along the coast. My best ideas often come to me while running.