Mathematical Experimentation and Sociological Theory: A Critical Analysis

The purpose of this paper is to suggest a serious problem with empirically based theory in sociology (low explained variance, on the average about 10 percent), a probable reason for that problem (traditional methodology) and a possible solution (mathematical experimentation and correlation). Data are reviewed from a number of mathematical experiments and correlational studies-of status, group learning, gambling, and voting. All of these involved psychophysical measurements and/or objective measurements at the ratio level. The resulting theoretical equations explain from 85 to 99 percent of the variance. Sociology as a scientific endeavor has made substantial progress since its inception around the turn of the century. Although there have been false starts, the equivalent of growing pains, sociologists and social scientists generally have learned to bring their theoretical and research endeavors together, to develop theory based on experimental and correlational data. Furthermore, if one attempts an inventory of research based theory in the social sciences, one is almost inundated. For example, the Berelson and Steiner (1964) inventory contains 1045 theoretical conclusions about how we are likely to behave in sex, marriage, religion, politics, business and other vital areas of life. While some of the theoretical propositions are better supported by data than others, a major criticism of the book is that the authors left out much of what could have been included. In other words, social scientists have, during the past several decades, focused upon discovering new empirical relationships which could be formulated into more or less general theoretical statements and have succeeded admirably in the task.