Forecasting in the social and natural sciences: An overview and analysis of isomorphisms

This article identifies and analyzes several points of similarity in the structure and context of forecasting in the social and natural sciences. These include: the limits of identities or universal laws as a basis for forecasts; the corresponding need for simplifying parametric representations of one or more of the variables that enter into identities; various sources of uncertainty about parameterizations; intrinsic limitations on predictability or forecasting accuracy in large-scale systems; the need for sensitivity analyses of model responses to changes in exogenous variables and/or parametric structures; problems of model linkage; and the social (organizational and political) context of forecasts. Suggestions for future lines of inquiry are made in each case. Several of these are such that they can benefit from a sharing of experience and expertise across disciplinary lines.The research reported herein was supported in part by the IC² Institute, The University of Texas at Austin, and in part by National Science Foundation Grant Number SES-8411702. However, the opinions expressed in the paper are those of the authors and do not necessarily reflect those of the sponsoring organizations. We appreciate the advice and comments of Jesse H. Ausubel, Robert S. Chen, Judith Jacobsen, and Richard C. Rockwell on earlier versions of this paper.The National Center for Atmospheric Research is sponsored by the National Science Foundation.