IDEA—a conversational, heuristic program for Inductive Data Exploration and Analysis
ACM '67 Proceedings of the 1967 22nd national conference
TRACE time-shared routines for analysis, classification and evaluation
AFIPS '67 (Spring) Proceedings of the April 18-20, 1967, spring joint computer conference
| Hi-index | 0.00 |
Social scientists are largely concerned with the discovery of patterns and relationships in multivariate data. The techniques used have been, for the most part, the standard tools of multivariate analysis---correlation, regression, factor analysis, and so forth. Unfortunately, the nature of these techniques has tended to impose implicit theoretical assumptions and constraints on the social sciences. The conceptualization of a variable as the weighted sum of other variables is methodologically unhealthy in the study of human beings and societies, in which it is precisely the complex interaction effects among variables which are of the most theoretical interest. In the social sciences it is especially prevalent, and especially significant, for the relationship between two variables to be dependent on the values of other variables---and for the nature of this dependence to be a function of still other variables. Traditional statistical methods are ill-suited for uncovering such hierarchies of interaction effects.