Statistics in Society: Problems Unsolved and Unformulated

Abstract Many problems important to society have major statistical components, yet are not wholly statistical in the usual sense. Examples of such problems are discussed, and a tentative classification is proposed. The first examples are those of inconsistencies and the like in large data sets, for example, census information showing a number of women, aged 15 through 19, with 12 or more children. The second group of examples deals with ambiguity of classification, for example, of ethnic origin. Other problems discussed briefly include clarity of statistical graphics, and confidentiality. Stress is laid on novel theoretical issues posed by these problems, and also on the social or political forces that generate the problems. A professional response is proposed in which resources are sought to study with care the behavior of relevant statistical activities.