Statistics in Business and Industry: The Path Forward
The ASA will celebrate its 175th anniversary in 2014. In preparation, column “175”—written by members of the ASA’s 175th Anniversary Steering Committee and other ASA members—will chronicle the theme chosen for the celebration, status of preparations, activities to take place, and, best yet, how you can get involved in propelling the ASA toward its bicentennial.
Gerald (Gerry) Hahn is a retired manager of statistics at the (current) GE Global Research Center, where he worked for 46 years. He is a co-author of four books and many articles, recipient of numerous awards, and Fellow of the ASA and American Society for Quality. He holds a doctorate from Rensselaer Polytechnic Institute.
Necip Doganaksoy is a principal technologist at the GE Global Research Center and an adjunct professor at the Union Graduate College School of Management in Schenectady, New York. He is a co-author of two books and a Fellow of the ASA and American Society for Quality. He holds a doctorate from Union College.
Happy (forthcoming) 175th birthday, ASA! This is a good time to reflect briefly on our past and to consider where we need to be heading. Our emphasis will be on statistics in business and industry—the focus of our combined 80 years as statisticians (spanning more than a third of the life of the ASA)—but we think our comments have general applicability. They reflect views expressed in our two recent books.
Statistics gained recognition in business and industry (beyond Bell Labs) in the years following World War II. Early applications were principally in manufacturing and (what was then called) quality control—building on the work of Walter Shewhart, W. Edwards Deming, and others. Data analyses were left mostly to statisticians in light of the complex calculations required using early computing equipment and the newness of the field. Over time, statisticians became heavily involved in “fire-fighting” projects, such as combating defective manufactured product and premature field failures.
Today, we live in a vastly different world, largely as a result of advances in computer technology and statistical methodology. Many industries—from semi-conductors to pharmaceuticals—rely heavily on statistics in their operations. Statistics also is making key contributions to advancements in the biological sciences, to cite just one example.
Moreover, “user-friendly” statistical software is readily accessible. Globalization has resulted in the outsourcing of many statistical analyses. As a result, statistics is used in business and industry more extensively than ever before.
However, much of the work is performed by practitioners, rather than professional statisticians. “Weibull analysis” is part of the toolkit for many engineers working in reliability. Mixture experiments are widely used by chemists to optimize product formulations. Manufacturing engineers rely on statistical studies to validate measurement systems and design experiments to improve product performance. Only the pharmaceutical industry—with its statutory need for rigorous product assessment—appears to be a, perhaps partial, exception to this extensive democratization of statistics.
These developments present fresh opportunities for statisticians. Relief from more routine data analysis has opened new dimensions for us. The changing environment allows us to play a more prominent role—even as we continue to encourage and help our nonstatistical colleagues use statistics effectively. This involves exerting statistical leadership in seeking out and addressing the key quantitative challenges that face a business and, in general, playing a proactive role in averting problems, rather than responding to them after they occur. It calls for seeking applications in continuingly broader domains and integrating statistical concepts directly into addressing often complex and unstructured business problems. This has resulted in what Roger Hoerl and Ronald Snee call “statistical engineering.”
In such expanded roles, statisticians need to become increasingly and intimately involved, often as team members and sometimes as leaders, in such areas as the following:
Getting good data up front. Traditionally, we have focused principally on data analysis, often requiring extensive initial data cleaning. The resulting analyses and data mining are only as good as the data upon which they are based. Much more emphasis is needed on getting the right data in the first place. This frequently calls for statistically designed experiments, but often goes beyond. The planning of comprehensive test programs during product design to assess and help improve performance and reliability is just one example.
The development of early-warning systems. Statistical monitors that process the data as generated are increasingly embedded into business processes. Applications range from systems that continuously track the performance of, say, an automobile or locomotive to signal impending failures to ones that examine ongoing financial transactions to provide early identification of vulnerable accounts.
Systems integration. This may involve dynamically combining data from multiple sources to achieve creative solutions to large-scale problems. A typical example is optimizing jet engine maintenance scheduling—to maximize reliability with minimal adverse effect on operations—by combining the results of recent inspections with information on in-flight performance and the integration of additional data from service shops, parts availability records, and flight schedules. Such integration often calls for the marriage of statistical and nonstatistical approaches, as, for example, in the merger of statistical and engineering process control into a single system that leads to both short-term and long-term quality improvement.
Exciting new challenges for statisticians continue to emerge. These include leveraging online information (e.g., early identification of consumer preferences by searching social media); the making of further inroads into finance (by developing improved risk models); personalized medicine (to find patient groups that are uniquely responsive to a particular treatment); and such hot areas as bioinformatics, climate change, nanotechnology, and improved customer servicing. We anticipate, moreover, a particular acceleration of opportunities for statisticians outside North America, especially in developing countries.
There is, in addition, much interest in emerging applications under such banners as Big Data, cloud computing, and analytics. These revolutionize the way organizations capture, store, share, and use large volumes of data. We need to be on top of these areas to leverage new opportunities that will arise as advances in technology make massive and timely computations still easier and even more accessible. This will call for even greater use of data-intensive methods in general and such approaches as Bayesian methods and incisive statistical graphics in particular.
Finally, broadening our horizons, we see future statisticians from all application areas serving increasingly as citizens of the world. Paraphrasing former U.S. Secretary of State Madeleine Albright, statistics are in the center of many arguments. We need to work toward expanding the role of statistical literacy to developing routine mechanisms for critically and impartially examining the numerous data-based claims—often from questionable data and/or analyses—with which our society is constantly bombarded.
Needless to say, all the preceding requires us to communicate and interact more effectively than ever. The now familiar quote by Google Chief Economist Hal Varian about statistician being “the sexy job in the next 10 years” might seem a little dramatic, but it holds much truth. It is up to us to recognize the changing environment and leverage the new opportunities.
Hahn, G.J., and N. Doganaksoy. 2008. The role of statistics in business and industry. Hoboken, NJ: Wiley.
Hahn, G.J., and N. Doganaksoy. 2011. A career in statistics: Beyond the numbers. Hoboken, NJ: Wiley.
Hoerl, R.W., and R.D. Snee. 2010. Moving the statistics profession forward to the next level. The American Statistician 64(1):10–14.
Jensen, W. (Editor). 2012. Statistics to facilitate innovation: A panel discussion. Quality Engineering 24(1):2–19.
Kettenring, J.R. 2012. Statistics research at Bell Labs in the regulated monopoly era. International Statistical Review 80(2):205–218.
Steinberg, D.M. (Editor). 2008. The future of industrial statistics: A panel discussion. Technometrics 50(2):103–127.