Abstract

This paper summarizes and extends the arguments in a number of earlier papers [2],[3],[4],[5],[6]. Although it is meant as a contribution to the methodology of the social and behavioral sciences, I think my argument actually applies to all disciplines that use statistics. The common concern in the papers and chapters mentioned above is to demarcate the responsibilities of the statistician and those of the empirical scientist. This means we assume that there is a legitimate academic discipline called "Statistics". This is, by no means, uncontroversial. Many scientists feel that they do not need statisticians to analyze their data, and many university administrators think that statistics is just an undergraduate course that students take to satisfy the general quantitative requirements. Quite a few statistics departments have disappeared, or could easily disappear, because it is tempting to distribute statisticians over the quantitative programs of various disciplines. In order to describe what belongs to science and what belongs to statistics I have to grope around in the murky area called the Foundations of Statistics.

Citations