Home » Additional Features, ASA Leaders Reminisce, Featured

ASA Leaders Reminisce: Ray Waller

1 March 2016 263 views 3 Comments
Jim Cochran
    In the 15th installment of the Amstat News series of interviews with ASA presidents and executive directors, we feature a discussion with 1995–2001 executive director, Ray A. Waller.

    Ray Waller

    Ray A. Waller earned his BA in mathematics from Southwestern College in 1959 and his MS in statistics from Kansas State University in 1963. Upon graduating with his PhD in mathematical statistics with a minor in operations research from The Johns Hopkins University in 1967, Waller joined the faculty of Kansas State University, where he taught undergraduate and graduate statistics courses, provided statistical consulting services across the university, and directed the theses of several MS and PhD students.

    In 1974, Waller moved to the Los Alamos National Laboratory’s Statistics Group, where he ultimately became the directorate office leader for university research and science education. In 1995, he became executive director and secretary of the ASA.

    Throughout his career, Waller has built and led several successful teams of statisticians, systems analysts, engineers, economists, education professionals, and scientists in collaborative interdisciplinary research projects involving innovative applications of statistics. He has authored or coauthored three books and coedited two volumes. Also, his research has been published in several academic journals, including the Journal of the American Statistical Association, Technometrics, and Annals of Statistical Mathematics.

    Q: You and David Duncan coauthored “A Bayes Rule for the Symmetric Multiple Comparisons Problem.” This article was published by the Journal of the American Statistical Association in December of 1969, at a time when Bayesian statistics and Bayesian statisticians had not been fully embraced by the statistics community. What was the reaction of the statistics community to this paper at the time it was published?

    A: The paper presents the Bayesian methodology and tables of critical values for the symmetric multiple comparisons (MC) problem. It was based on my dissertation completed in 1967 with David Duncan as my major professor. The symmetry assumption is that the distributions under study have equal variances.

    Duncan was well known in multiple comparison analyses from the wide use of his multiple-range method in research published in agriculture and other applied journals. He began researching Bayesian methods for multiple comparisons in the late 1950s. He believed the results in my dissertation extended his results to provide a Bayes rule for the symmetric MC problem. Further, he thought the method was superior to his widely used multiple-range test. The methodology became more user-friendly in 1975 when a Kansas State University colleague, Ken Kemp, rewrote my FORTRAN code to be a much more efficient algorithm.

    A paper published by Waller and Kemp in the Journal of Statistical Computation and Simulation in 1975 discussed the algorithm. Many copies of the algorithm were distributed, including a copy to SAS. SAS added the Waller-Duncan Bayes procedure as an option for completing multiple-comparisons analyses.

    You are right in saying that Bayesian statistical methods have gained favor in the last 50 years, including the formation of the Bayesian Section of the ASA. I consider David Duncan’s missionary work among his consulting contacts and the numerous users of his multiple-range test over the years as important factors in the acceptance, use, and frequent citation of the 1969 JASA article over the last 30+ years. Also, the acceptance of Bayesian MC analysis as a legitimate statistical method by various applied journals and its availability on SAS were important factors in its usage.

    Q: Did any particular statistical project or application motivate you and Vincent T. Covello to edit and publish Low-Probability High-Consequence Risk Analysis: Issues, Methods, and Case Studies in 1984?

    A: The Statistics Group at the Los Alamos Scientific Laboratory—now the Los Alamos National Laboratory, or LANL—began reliability and risk studies/analyses prior to my arrival on staff in 1974. The U.S. Atomic Energy Commission—a forerunner to the U.S. Department of Energy formed in 1977—and the Nuclear Regulatory Commission were interested in estimating the reliability of safety systems and conducting risk analyses for nuclear power plants. This was prior to accidents occurring in two operating nuclear power plants—Three Mile Island in Pennsylvania and Chernobyl in the USSR.

    The goal in operating nuclear power plants is “no failures” (i.e., low probability of failure). On the other hand, any failure of a nuclear power plant has the potential for great damage and heavy losses (i.e., high consequences). An operational goal of no failures results in little, if any, data for analyzing nuclear power plant operations. Thus, new and/or improved analytical, statistical, and reliability estimation methods were being studied and developed.

    The Society of Risk Analysis, U.S. Nuclear Regulatory Commission, U.S. Department of Energy, and U.S. Environmental Protection Agency jointly sponsored The Society for Risk Analysis International Workshop on Low-Probability High-Consequence Risk Analysis: Issues, Methods, and Case Studies in June of 1982.

    The Society for Risk Analysis started a book series in 1983 called Advances in Risk Analysis. Eight volumes were published between 1983 and 1991. Vincent Covello was the lead editor for volume one, The Analysis of Actual Versus Perceived Risks. The edited papers from the workshop described above were published as volume two, Low Probability/High Consequence Risk Analysis. Vincent Covello and I served as co-editors of volume two. This work provided preliminary background for reliability and risk analysis research work performed by the statistics group at LANL during my tenure in the group.

    Q: How did your experience at the Los Alamos National Laboratory satisfy your intellectual curiosity and your need to grow as a statistician?

    A: As a consultant at the White Sands Missile Range in New Mexico starting in 1968 and as a summer employee there in 1971, I interacted with a variety of statisticians, engineers, and scientists studying the performance and reliability of missile systems. The experiences and professional challenges of being part of an interdisciplinary scientific research team proved to be a stimulating and productive work environment for me. I found great satisfaction in researching solutions to real-world problems; communicating statistical and other analytical results to researchers, government regulators, and managers; and seeing the opportunity to impact changes in polices and practices with the potential to produce improved end results.

    While considering a sabbatical to conduct research in reliability and risk analysis, I was invited to interview for a position with the statistics group at LANL. After visiting LANL and finding an opportunity to be part of reliability and risk analyses for safety systems of nuclear power plants, I decided to make a career change, rather than take a sabbatical leave.

    Early in my time at Los Alamos, I met and began collaborations with Harry Martz, a statistician at Texas Tech University doing reliability research. Harry soon became a colleague at LANL and we ultimately coauthored the book Bayesian Reliability Analysis in 1982. We also published several technical reports and articles together, and we jointly presented short courses on reliability and risk analysis methodology. The publications and short courses were sponsored by programmatic research support from the U.S. Department of Energy and U.S. Nuclear Regulatory Commission.

    The work at LANL evolved into a career, offering the multidisciplinary approach to real-world applications of statistical/analytical analyses that I had found rewarding during my consulting and summer services at White Sands. Finally, I believe my research and management experience at LANL played an important role in my having the opportunity to serve as the ASA executive director.

    I am very grateful to professors, mentors, and numerous colleagues for the three different and challenging career opportunities: 1) statistics instruction and research in colleges and universities; 2) research and management at LANL; and 3) ASA administration.

    In addition to gratitude to the various individuals, I feel strongly and have stated many times to colleagues and program funders that many important decisions have been and will continue to be made without benefit of available statistical analyses, but many, if not most, can and will be better decisions with appropriate statistical involvement and considerations.

    Q: You and your wife, Carolyn, established the Waller Education Award in 2002 and the Waller Distinguished Teaching Career Award in 2014. What experiences motivated you and Carolyn to establish these two awards?

    A: The ASA was my primary professional association beginning with my student membership in 1963. It was helpful to me through my careers in academia, a government laboratory, and the ASA.

    I began teaching a first course in statistics as a graduate teaching assistant at Kansas State University in 1961, and I continued to teach a similar course as either an instructor or professor at the Evening College of The Johns Hopkins University, Towson State College, and Kansas State University through August 1974.

    I believe the first statistics course is an important requirement for many majors and is well taught by a capable and creative group of instructors and professors at all levels of learning, from high school through college. Further, I have had the privilege of meeting and talking with many of those dedicated individuals over the years.

    When Carolyn and I began discussing possible ways to honor people teaching that first course in statistics, it seemed natural to select honorees from teaching assistants/instructors/professors early in their careers. Working with ASA Associate Executive Director and Director of Operations Steve Porzio, guidelines and endowment funding for the Waller Education Award were finalized.

    We are pleased with and grateful for the professional manner in which the ASA Statistical Education Section has managed and administered the Waller Education Award since its inception. After about 10 years of experience with the Waller Education Award, the Statistical Education Section expressed an interest in developing an award recognizing contributions for more senior faculty members. The section drafted guidelines for review by the ASA staff, as well as by Carolyn and me. The guidelines were finalized and we provided funds to endow The Waller Distinguished Teaching Career Award starting in 2014.

    Q: What was most challenging and most gratifying about serving as executive director of the ASA?

    A: It was my privilege and joy to serve as ASA executive director after being a member for 30+ years. The most gratifying part of the position was to meet and work with many dedicated volunteer ASA officers and leaders, the competent ASA office staff, and the ASA membership to continue providing quality support for publications, professional meetings, and continuing education.

    It was gratifying to see office professionals in computing, finance, education, publications, and meetings and professional statisticians in academia, government, and industry develop mutual respect for the different contributions each makes toward accomplishing the ASA goals set by the board. The talents and abilities of both members and staff are necessary to deliver a wide variety of services and products to the membership and the public.

    One challenge we faced together was the installation of a new computing system for the ASA. Soon after a system customized for ASA office operations was installed and operating, it was determined to be inefficient in operations and costly to maintain. Director of Information Technology Tim Gill and Steve Porzio led the ASA effort to select and install a new computing system composed mostly, if not entirely, of off-the-shelf components. The resulting system was installed successfully and resulted in more efficient and less costly operations.

    To me, the preceding experience is similar to many other challenges I faced as ASA executive director and that we all face in our lives and careers—a new career problem/assignment may present challenges and frustrations while you are pursuing a solution, but it can become a great source of gratification upon successful completion.

    I will always remain grateful for the opportunity to serve as ASA executive director and for the opportunities it provided me to know and to work with talented, dedicated, and capable ASA members and staff.

    1 Star2 Stars3 Stars4 Stars5 Stars (No Ratings Yet)
    Loading...

    3 Comments »

    • New Exploitation Blog said:

      Reminisce Magazine 4 Issues

      […] d his results to provide a Bayes rule for the symmetric MC problem. Further, he […]

    • New Process Blog said:

      Reminisce Magazine 4 Issues

      […] rsity colleague, Ken Kemp, rewrote my FORTRAN code to be a much more efficient a […]

    • New Inspiration Blog said:

      Reminisce Magazine 4 Issues

      […] e began researching Bayesian methods for multiple comparisons in the late 1950s. […]