Quality and Relevance in the First Statistics Course

Abstract

Professor Cox is certainly right in stressing the importance of the first course in statistics. Many future national leaders and citizens form their attitudes toward the subject here, attitudes that will later encourage or impede widespread use of statistics as a national resource. The first course is also a key recruiting ground for future statisticians among students who typically have little awareness of the discipline prior to being asked to take an introductory course. If statistics has a role in national prosperity, that role begins in the first course for general students. The quality and relevance of an introductory course depend on effective integration of technologyand pedagogywith content. Moore (1997) gives general background, with comments by a distinguished group of discussants. 2 Technology and Pedagogy It is almost essential that a relevant introduction to statistics incorporate use of statistical software, both to allow work with substantial real data and to form good habits of practice. Modern content greatly increases the need for adequate computing. This is true even if we avoid contemporary topics such as resampling methods and Bayesian inference. Effective exploratory analysis of data is in practice only feasible when graphics and calculations are automated. Even inference is in practice characterized by back-and-forth movement between data and models, with models providing a basis for inference and the data allowed to criticize and even falsify models via diagnostics. The modes of thinking needed for working statistics are quite different from the “derivations down from models” mode of mathematical statistics. These modes of thinking, and the tools that implement them, are best learned in the context of actually working with data. In the absence of adequate computing, a first course in statistics is in the nature of “dry swimming”. Comprehensive use of technology in courses for large groups of students is expensive. This explains why in the United States only 61% of elementary statistics courses taught by Ph.D.granting departments of statistics use computing (Lutzer et al., 2002). The corresponding percent for mathematics departments is only 48%. It appears that even many statistics departments disagree with the priority assigned by Professor Cox. In developing nations, availability of resources rather than the will to use them is a barrier. Park Chan-mo, president of Pohang University of Science and Technology, comments that most science programs in South Korea “tend to focus on book learning