Forecasting in social science research: imperatives and pitfalls

Abstract

Social science research typically investigates society's evolving economic and social conditions and the processes shaping them. The theories we develop or refine to explain such conditions and processes are often tested retrospectively. One approach is to apply simulation models, grounded in our theories, to previously known conditions and project them forward to assess how well they predict present circumstance. Somewhat less frequently, research topics look beyond the present to identify society's looming conditions, their associated problems and opportunities, and possible need for public regulation or control. This task requires not just a sound conceptualization of contemporary processes, but also an understanding of how they are likely to change during the forecast period. We know, for example, that the relative importance of different component variables in a system will almost certainly rise and fall, while others will be added or deleted, and the patterns of causality between them will reconfigure. These, in turn, reflect technological advance; shifting supply and demand relationships in resources, goods and services; and human perceptions about needs and wants. Unsurprisingly, forecasting is the more difficult of the two tasks because it combines sound knowledge of current circumstance and process with informed, but nevertheless speculative, analysis about their future trajectories.