Forecasting errors: The importance of the decision-making context

Abstract

This article demonstrates the importance of context on forecast errors. It describes the development of an unrealistic transportation planning forecast for Denver, Colorado, noting its influence on air quality planning for Denver. The importance of two contextual features, procedural rationality and conflict of interest, to the introduction of error is discussed. The case described implies that different decision-making contexts define different rules of rationality, and substantive rationality may not fit the rules. The type of context and not necessarily a code of ethics will determine this."As for the Future, your task is not to forsee, but to enable it." (Antoine de Saint-Exup~ry)Investigations of forecasting errors generally assume that the forecasts have been developed using the best scientific information. Their analysis therefore revolves around the accuracy and uncertainty of model predictions. They concentrate on issues of verification and technique (Asher, 1979;Brewer, 1973; Makridakis, 1982; Anthes and Baumheftner, 1984;. Using the same assumptions, psychologists investigate sources of error attributable to cognitive limitations, judgmental biases and unexamined influences of judgment (see Stewart's chapter; Hogarth and Makridakis, 1981;Mumpower and Dennis, 1979). Such investigations assume that thought and action, facts and values can be separated.There are, however, many technical forecasts of interest that are developed not purely as efforts toward compiling scientifically accurate information, but rather as integral parts of policy making. Their purpose is not just to present facts but to influence and affect decisions. Their development and use as instruments of persuasion is of critical importance in contexts of conflicting interest and bureaucratic decision-making.In policy making as elsewhere~ scientists are not expected to conjure up forecasts and proposals free of controversy: credibility of arguments and numbers is assumed to be important to the persuasiveness and ability to influence policy. Policy makers must be able to assert that forecasts supporting their decisions rest on the best information available. A degree of substantive rationality is assumed to reign.There are decision-making contexts, however, in which it is more important to be persuasive than to be right or wrong. Political scientists point out that in the absence Climatic Change 11 (1987) 81-96. 9