Towards An Understanding Of Statistical Thinking

The Challenge of Developing Statistical Literacy, Reasoning, and Thinking
Pfannkuch, M. & Wild, C.
Ben-Zvi, D. & Garfield, J.
Kluwer Academic Publishers, Dordrecht, The Netherlands

There has been an increasingly strong call from practicing statisticians for statistical education to focus more on statistical thinking (e.g., Bailar, 1988; Snee, 1993; Moore, 1998). They maintain that the traditional approach of teaching, which has focused on the development of skills, has failed to produce an ability to think statistically: "Typically people learn methods, but not how to apply them or how to interpret the results" (Mallows, 1998, p. 2).<br>Solutions offered for changing this situation include employing a greater variety of learning methods at undergraduate level and compelling students to experience statistical thinking by dealing with real-world problems and issues. A major obstacle, as Bailar (1988) points out, is teacher inexperience. We believe this is greatly compounded by the lack of an articulated, coherent body of knowledge on statistical thinking that limits the pedagogical effectiveness even of teachers who are experienced statisticians. Mallows (1998) based his 1997 Fisher Memorial lecture on the need for effort to be put into developing a theory for understanding how to think about applied statistics, since the enunciation of these principles would be useful for teaching.<br>This chapter focuses on thinking in statistics rather than probability. Although statistics as a discipline uses mathematics and probability, as Moore (1992b) states, probability is a field of mathematics, whereas statistics is not. Statistics did not originate within mathematics. It is a unified logic of empirical science that has largely developed as a new discipline since the beginning of the 20th century. We will follow the origins of statistical thinking through to an explication of what we currently understand to be statistical thinking from the writings of statisticians and statistics educationists.

The CAUSE Research Group is supported in part by a member initiative grant from the American Statistical Association’s Section on Statistics and Data Science Education