Decomposing Forecasting

Abstract

The act of forecasting one’s behavior or performance is both commonplace and consequential, but it is also difficult. Previous research has identified a host of systematic forecasting errors. We suggest that existing findings can be better synthesized, and future research can proceed in a less piecemeal fashion, through the introduction of a general model that describes how forecasts unfold. In our salience-assessment-weighting (SAW) model, we outline three steps that describe how people translate information at their disposal into an accurate forecast of a future outcome. Dimensions potentially relevant to the outcome become salient; one’s standing on that dimension must be accurately assessed; and one must appropriately weight the importance of that dimension to translate it into a forecast. We illustrate how this SAW model is helpful in unifying previous research findings, identifying how and when forecasts go astray, and suggesting questions for future research.