Criteria For Critiquing Social Science Computer Models With Application To Barr-Gale Model For Consumer Price Index Of Food

The authors propose six questions relevant to the evaluation of all social system models, and apply them to the Barr-Gale model for forecasting the Consumer Price Index for Food. While analysts invest much effort in developing new economic models, relatively little attention is paid to critiquing existing models. For models to become scientifically acceptable and widely used tools for policy design in the social sciences, they must be accompanied by complete documentation and data. Then independent investigators can confirm published results and determine a model's sensitivity to changes in assumptions. Greater emphasis on critical analysis and standard procedures for evaluation will help potential users evaluate models for accuracy and suitability for their purposes.