Microcomputers and Improving Social Science Prediction

Abstract

This article analyzes a type of microcomputer program that can process a set of (1) prior cases or incidents, (2) predictive criteria for separating the cases into winners and losers, or two or more other categories, and (3) relations between each prior case and criterion in order to arrive at a decision rule that is accurate and makes substantive sense. Micro computers allow the user to change quickly the various inputs to see the effects the changes have in reducing inconsistencies in applying the decision rule to the data set. This is educated reiterative guessing, as contrasted to more mechanical least squares approaches.