Classical Statistical Theory

Abstract

In an area where there is so much acknowledged turmoil, such outspoken disagreement, and in which so many divergent points of view are so strongly represented as in statistics today, it is almost anomalous to refer to one particular theory as the ‘classical’ one. It is the stranger to use this term, in that the theory to which it is applied may be said to be no more than fifty years old. But in statistics, as in mathematics, things happen quickly — far more quickly than the layman, or even the professional consumer of statistics or mathematics, realizes. Many of the fundamental ideas of the theory of testing statistical hypotheses were developed in the first quarter of this century by Karl Pearson, Jerzy Neyman, and R. A. Fisher. By the nineteen-forties, the view, primarily associated with the name of Neyman, that the fundamental form of statistical inference was the choice between statistical hypotheses, had come to dominate other views in the English speaking world. It merely dominated other views among statisticians; but it utterly overwhelmed other views among those whose interest in statistics was primarily practical.