Applying Knowledge and Confidence Information to Predict Achievement in Forecasting

Abstract

In this paper, we explore the relationship between confidence and basic knowledge information collected prior to a forecast and forecast performance (i.e. achievement). Forecasters were asked at the beginning of the 2008 Major League Baseball season to predict end-of-year performance in five statistical categories for both hitters and pitchers. At the time of the initial forecast, participants were asked to rate the confidence in their forecast (0-100 scale) and were asked simple knowledge-based questions related to the player they were making forecasts about (e.g. player position and team). A regression analysis was used to compare the participant’s performance on the statistical categories with their knowledge and confidence. In general, there was a significant relationship between performance in the statistical categories and both knowledge and confidence when compared individually. However, given both confidence and knowledge, confidence provides no additional significant information.