Jim Cochran

In the seventh installment of the Amstat News series of interviews with ASA presidents and executive directors, we feature a discussion with 2001–2007 ASA Executive Director William B. (Bill) Smith.

William B. Smith

Q: Bill, thank you for taking time for this interview. You earned your BS and MS degrees in mathematics from Lamar University and Texas A&M University, respectively, before earning your PhD in statistics from Texas A&M University. Did any courses that you took in pursuit of your mathematics degrees motivate your future studies in statistics?

A: In 1964–66, graduate statistics and computer science programs were still rare. My background had been ‘pure’ mathematics, focusing on analysis. I studied a wide range of mathematical areas, but real and complex variables as well as numerical analysis interested me most. While the complex variables side was not terribly important to my statistics training, the real analysis/measure theory was very helpful. The most important part my mathematical background was acquiring an understanding of what constituted a ‘proof.’

Q:You have worked extensively with a diverse set of organizations, including the Texas Agricultural Experiment Station, Instituto Nacional de Tecnologia Agropecuaria (INSTA, Argentina), Lower Colorado River Authority (LCRA), Scott and White Hospital and Clinic, and Texas A&M University Medical College. What particular projects were most interesting or challenging?

A: Each of these projects was interesting and challenging. Ag Station projects ranged from minimized bruising when harvesting peaches to dietary studies on race horses (see Growth Curves question). Renal and ophthalmology repeated measures studies with the medical groups; teaching multivariate and time series at INSTA; and examining corrosion on wall panels at a coal-fired power plant for LCRA. Usually accepted statistical practices and methods were adapted to fit the problems posed. For example, the LCRA project used an adaptive sampling procedure to examine randomly selected external wall panels, using both an indicator variable for the presence/absence of significant corrosion and continuous measurements of the extent or the corrosion.

Q:Relatively early in your career you worked with H. O. Hartley on a pair of projects for NASA. What impact did those experiences have on your career in statistics?

A: HOH was a wonderful mentor and friend. He was totally into his research and consulting projects, and our NASA collaboration was no exception. At the time of these grants, spaceflight was in its infancy with lots of data gathering and interpretation issues, including astronaut health monitoring, LANDSAT imagery, etc. These projects were ones with Big Data for the times, missing and partial records, and thus were multivariate by their very nature. I have spent most of my career with these types of problems.

Q: You and Anant Kshirsagar coauthored the book Growth Curves. What specifically motivated you and Kshirsagar to write this book?

A: Anant Kshirsagar and I have been fast friends for many years. His expertise in multivariate analysis is unsurpassed, so I could not refuse when he approached me with this project. Applications of multivariate methods, and particularly repeated measures experiments, had come to me from many venues, including as examples from salary discrimination studies and USDA food inspection quality control instruction to research projects on dietary studies for racing quarter horses and ocular drug studies with veterinary ophthalmologists. So, in the book, I provided some chapters and the applied/experimental input, Kshirsagar the theoretical.

Q:You spent more than 40 years as a faculty member with the statistics department at Texas A&M University (TAMU), during which time you were recognized as a TAMU Outstanding Teacher. You then moved into administrative roles, first at TAMU and then the NSF and ASA. What motivated you to move from teaching into administration? What do you miss about teaching?

A: I still teach in the graduate program at TAMU, and I did so while doing TAMU administration. I am now involved with the TAMU analytics MS degree. My administrative work began some 42 years ago when the statistics department successfully initiated an undergraduate mathematical sciences degree with the mathematics department. The college of science dean noticed my involvement in this lengthy degree-approval process and asked me to serve as his assistant dean. This invitation led me into coordinating research efforts through the college and supervising the construction of a science laboratory and lecture hall building, as well as continuing my work in the statistics department. By the way, my campus construction experience aided greatly in the ASA headquarters building project.

Q: What recent challenges do you feel the ASA and our discipline have successfully confronted? What challenges do you see in the near future?

A: In recent years, the ASA successfully met the fiscal challenges to the organization caused by the impact of open access and virtually freely available publications. It has created a certification process, bought and renovated a headquarters building, improved continuing education efforts both at JSM and online, organized and conducted smaller, more focused meetings, and improved JSM. None of this would have happened without active membership input and participation—the membership, ASA leadership, and ASA staff can give each a collective pat on the back for these successes. Some present and future challenges include the following:

• Big Data
• Data mining by statistically untrained computer scientists
• Decisions by industry and government using poor information and analyses

I think the ASA is in good hands—both the board and ASA staff are open minded and dedicated to the society. I wish for them the same support I had at the ASA.

Please return to this column next month, when we will feature an interview with 2010 ASA President Sastry G. Pantula.