Discussion

Extract

The papers given at this symposium have been directed to two problems:

1. (1) the needs of the physical sciences which the social sciences should fulfill, and

2. (2) the capacity of contemporary social science to satisfy these needs.

Consideration of the first problem divided itself into two parts: (a) the needs involved in the process of answering questions in the physical sciences, and (b) the needs involving application of information gained by questions in physical science. The role of social science with respect to (b) is generally recognized, but the role with respect to (a) is not. Many, in science and philosophy, still hold to the notion of a non-evaluative, non-normative, purely objective, or what-have-you physical science. Professor Littauer, however, raised a critical problem in this regard: on what basis is the decision made to accept or reject a scientific hypothesis? We are in a position today to assert that such decisions do involve an evaluative judgment. Recent developments in mathematical statistics, as represented in Wald's On the Principles of Statistical Inference, indicate that the selection of the best statistical testing procedure involves at least a knowledge of the social consequences for the various mistakes one might make in using any specific testing procedure. The introduction of the “weight function” into the method for selecting among alternative hypotheses points in a striking way to one door through which knowledge and evaluation of possible social phenomena enters physical science. The social scientist is badly needed here. It is true that today this aspect of experimental method is seldom self-consciously faced. Where it is, the evaluations are generally intuitive and are not based on sound sociological knowledge. But it is becoming increasingly clear that the continued development of experimental methods in all sciences depends on our reducing the question of what constitutes a “best” test to experimental terms.