Statistical Thinking in Policymaking
Guy Caruso, Mary Foulkes, Mary Gray, David Salsburg, Milo Schield, and Ronald Snee
Amstat News is inviting articles for a new series called “Statistics for Policymakers” based on the broad interest in a January blog entry on the ASA Community called “Statistics for Future Presidents.” The blog entry, inspired by the book Physics for Future Presidents, invited ASA members to share what they think policymakers should know about statistics. Please send your column ideas to ASA Director of Science Policy Steve Pierson at email@example.com.
Statistical thinking and methods are an integral part of modern scientific activity and problemsolving. Going beyond scientific investigations, statistical thinking has been successfully applied to problems in military tactics, manufacturing, economic analyses, and marketing. Statistical thinking provides the strategy needed to guide the tactics of statistical modeling. The power of statistical concepts, methods, and tools lies in their ability to identify problems in clear, unambiguous ways and determine what is known, what needs to be known, and what can and cannot be known. Identification and structuring of problems is critical to successful problemsolving.
Can statistical thinking be applied more effectively to policy decisions in government? Those of us who have engaged in using statistical thinking and developing new methods to solve problems frequently meet individuals who are arguing over unimportant or unknowable aspects of a problem. A little statistical modeling, guided by statistical thinking, is often able to clarify these problems and lead to solutions.
Much of the current political discussion over immigration, the national debt, health care, regulation of banks, etc. often sounds like such a situation. Policymaking and decisions lead to problems that are large, complex, and unstructured. Statistical engineering as discussed by Roger Hoerl in “The World Is Calling; Should We Answer?” published in the May issue of Amstat News, provides guidance for using statistical thinking to address such problems.
The statistical approach leads one to first identify and structure the questions that need to be answered. Are they answerable? Are there data available to answer them? How can new data be derived if needed? What is a reasonable range of uncertainty that we can expect or afford with those answers? What is the maximum level of uncertainty that can be accepted before acting on those answers?
We, thus, challenge the statistical community and those involved in policymaking: How can statistical thinking and modeling be used to aid in policy decisions in government? Are there specific problems that can be addressed with this approach? Send us your proposal for an article in Amstat News about this topic.