Marie Davidian is currently the President of the American Statistical Association. She received her Ph.D. in Statistics from the Department of Statistics at the University of North Carolina at Chapel Hill in 1987 under the direction of Raymond J. Carroll and is currently William Neal Reynolds Professor at Department of Statistics there as well as Adjunct Professor of Biostatistics and Bioinformatics at Duke University. Her interests include statistical models and methods for analysis of longitudinal data, especially nonlinear mixed effects models; methods for handling missing and mismeasured data; methods for analysis of clinical trials and observational studies, including approaches for drawing causal inferences; pharmacokinetic and pharmacodynamic analysis; combining mechanistic mathematical and statistical modeling of disease progression to design treatment strategies and clinical trials; and statistical methods for estimating optimal treatment strategies from data.
Statistics Views talked to Professor Davidian during the Joint Statistical Meetings last month in Montreal about her role as President of the American Statistical Association, her involvement with the International Year of Statistics, her previous presidency of the ENAR region of the International Biometric Society, her current research and teaching biostatistics.
1. You are currently William Neal Reynolds Professor at Department of Statistics of North Carolina State University as well as Adjunct Professor of Biostatistics and Bioinformatics at Duke University, and you spend several days a month at Duke Clinical Research Institute collaborating with clinicians and biostatisticians there on problems in cardiovascular disease research. Over the years, how has your teaching, consulting, and research motivated and influenced each other? Do you continue to get research ideas from statistics and incorporate your ideas into your teaching? Where do you get inspiration for your research projects and books?
Absolutely. I have to admit I am not doing so much work at Duke Clinical Research Institute this year whilst I am President. To me, developing new statistical methods cannot be done in a vacuum – you have to have a connection with real issues that arise in real application areas to motivate you and for you to understand what the problems are. Getting involved in a domain science like cardiovascular disease has been for me the motivating force for what I work on, like I mentioned earlier with the optimal treatment strategies. Over the years, just talking to the statisticians and clinicians at Duke works out as they will come with a problem in terms of a particular study and its analysis, which presents a challenging issue that involves missing or partial information and there is no real methodology to help deal with it. That has been the driving force for a lot of research I have seen and in the dissertations of my students too.
I actually have a NIH training grant which supports five pre-doctoral students. They take all their coursework at NC State and they go one day a week to Duke where they are placed in project teams. Each starts out as a low-level statistician working on data cleaning and simple analyses but as they progress, they become more integrated into the actual project to the point where they eventually become the statistician on a data analysis project there, and this has been the case for some of my senior students. Part of the goal is to actually have them see this whole process whereby you collaborate in a domain science area, problems arise in which new statistical methodological development is required and recognising what those are and what methodology is needed and then carry out the development of that methodology. I have had this grant since 2006 and the dissertation research of my students has come directly from something they encountered in a project team they were on. It’s great for them to be able to see that whole process.
To me, developing new statistical methods cannot be done in a vacuum – you have to have a connection with real issues that arise in real application areas to motivate you and for you to understand what the problems are.
2. As your presidency will draw to a close at the end of this year before handing over to Nathaniel Schenker, do you have any remaining priorities/objectives?
One of the biggest challenges the statistical community faces is defining our role within the area known as data science. This is also one of our greatest opportunities. As I noted in my presidential address and in some of my Amstat News columns, statistics and statisticians are often missing from the ongoing discourse on data science and big data, which is disheartening given the critical importance of statistical principles and thinking to this area. In June, my fellow presidents Nat Schenker and Bob Rodriguez announced a three-pronged ASA initiative to junp start the engagement of the statistical community with the community of data scientists. We have already achieved one of our goals, to offer ASA continuing education in the analysis of unstructured text data at the 2014 JSM and Conference on Statistical Practice, an area of great importance in many “big data” applications but one that many statisticians know little about. We are in process of organizing meetings between ASA representatives and data science stakeholders in technology and business to gain understanding of their perspectives and to emphasize the importance of statistics. We are also organizing an ASA workgroup to make recommendations on how graduate curricula in statistics should be revised to reflect the emergence of data science and to ensure our students are trained so as to have the requisite skills. A top priority for me is to have these efforts developed and well underway by the end of my term as president.
A second key objective for me is to see even more progress on another of my presidential initiatives, to improve our stature among scientists in other disciplines and to see more statisticians engaged in the American Association for the Advancement of Science (AAAS). In late September, I and ASA staff met with both Alan Leshner, CEO of AAAS, and Marcia McNutt, the new editor-in-chief of Science, which is published by AAAS, along with several of her senior editors. The goal was to discuss enhancing the stature of statistics and seeing statistics playing a more significant and visible role in Science. We received a very positive and enthusiastic reception from both, including great interest from Dr McNutt and her editors in highlighting statistics in Science, improving the reviews of Science papers in regard to statistical analysis, and recruiting more statisticians to serve on the editorial board. I would be thrilled to see some concrete steps in these directions in the works by the end of this year.
3. How do you think the ASA has evolved over the years and adapted to the changing needs of the statistical community?
Like many professional societies, several decades ago, the ASA focused most of its activities and efforts inward, toward our discipline. While this is certainly a key part of an association like ASA’s charge, particularly given the impact of statistics on science and society, an association like ASA also has a responsibility to represent the interests of our discipline outwardly and to speak for our profession on matters of science and policy where statistical thinking is critical.
This has become increasing critical with the data revolution. ASA advocacy for the importance of statistics in policy, education, and science benefits our entire community through increased opportunities, exposure, and influence. Over the past decade in particular, as a result of the leadership of several of my predecessors and ASA Executive Directors, the ASA has embraced this role and has added a Director of Science Policy and Public Relations Coordinator who have done much to promote the importance of our discipline and to elevate our voice on policy matters. This is just one of many ways the ASA has responded to the changing needs of our discipline and our community.
4. You are also Executive Editor of Biometrics. What makes Biometrics different from other journals in the field?
Biometrics is the flagship journal of the International Biometric Society. The Society represents a broad range of interests, all connected with research on biological sciences, but everything from medical and health-related activity through environmental, ecology, wildlife. While there are many journals that focus on one slice of that, such as Biostatistics on the human health effect, or the Journal of Biological, Agricultural, and Environmental Statistics, which is also an IBS journal with the ASA, but we do it all, and that makes the journal unique. We also try very hard to ensure swift review times. I think as a profession we really need to do more on review times as a whole and with Biometrics, we really emphasise fast but quality reviews. We have around 100 Associate Editors, as we are so broad and we emphasise to them the need to handle papers quickly, fairly and comprehensively, returning the reports quickly to the author. We are one of the few journals that publish monthly updates of our review times on our website and provide a lot of information on our review statistics to authors so that they can actually see how well we do. A lot of journals either do not do that or do it very infrequently, so I think that really sets us apart from other journals in the field. All journals are committed to swift reviews but I just did the statistics this morning (date of interview, 8th August 2013). Our target is 6-8 weeks as a median and we are more or less there. Amongst statistics journals, I am proud of what we do.
5. Who should be reading the Journal and why?
Everybody, as we are so cross-cutting. I think we are one of the top-tiered journals in our profession focussing on everything biological. We attract papers from some of the top researchers in various areas of applications in statistics. You can be assured that the articles that appear in the Journal are amongst the most innovative that there are to be found.
We as a discipline have been rather timid and hesitant to engage in this data revolution. We have not yet engaged to the extent that I think we should. Some of us are, but collectively, we have not yet embraced this challenge to the extent we should.
6. What has been the most exciting development that you have worked on in statistics during your career?
Over the years, I have been very fortunate to work on a number of different projects. I spent a lot of time in my early career back in the 90s as was one of the people in statistics who got into pharmacokinetics and pharmacodynamics early on when there were still a lot of methodological challenges. The pharmacokineticists were basically statisticians as they were doing statistical modelling – combining it with the kind of compartmental modelling they do for describing how your body processes drugs. Statisticians had not really been involved so I was among the first wave of statisticians who got into that area and developed some of the methodology and refined it. So I am proud of that aspect of my life. I then gravitated away from that into other areas. I am really excited about optimal treatment strategies with the emphasis now on personalised medicine and comparative effectiveness research, which is concerned with the best way to use existing treatments.
7. 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 statistics during the next few years?
This is an exciting and challenging time for us with the explosion of Big Data and all sorts of new problems to work on. A lot of the traditional methods we use may not be straightforward to scale up to deal with massive data, and there are other issues such as computing and data storage. This whole so-called ‘data revolution’ will force us to rethink the way we do things and collaborate with scientists in other disciplines who may not have our statistical thinking skills but have other skills in data handling and processing, and this will lead to many exciting avenues in statistical research.
8. What do you see as the greatest challenges facing the profession of statistics in the coming years?
We as a discipline have been rather timid and hesitant to engage in this data revolution. We have not yet engaged to the extent that I think we should. Some of us are, but collectively, we have not yet embraced this challenge to the extent we should. I have been going around encouraging such engagement. The area called ‘data science’ may not be clear to some. Some say that data science is broader than statistics as it involves computing and data visualization, others agree with Nate Silver’s response here at JSM when questioned about data scientists, that it is a “sexed-up term for a statistician.” It will all fall out in the next few years and we need to be in the middle of it. We are the statistical thinkers, we are the original data people, this is what we do.
9. Are there people or events that have been influential in your career?
I would certainly credit Dave Harrington for peaking my interest in statistics; my dissertation advisor Ray Carroll had a profound influence on me as I basically learnt how to do research from him and the skills that I use to this day. The opportunity to get involved in pharmacokinetics was through my friend David Giltinan from graduate school who was at Merckwhen I was in my early years at NC State. He introduced me to pharmacokinetics and we ended up writing a book together. He is a good friend of mine to this day and I think that book was a turning point for me. I am very proud of that book. It is perhaps outdated now but at the time, I felt that we had provided a very useful resource.