Effects of reasons analysis on the accuracy of predicting basketball games

It is argued that analyzing and listing reasons can lead to poorer predicti ons because reasoners either access information inappropriate to the task o r have difficulty integrating the information they do bring to mind. To tes t this hypothesis, self-described basketball experts predicted the outcomes of actual basketball games in a national tournament. Half of the participa nts were asked to analyze and list reasons for their predictions before mak ing them, and half were told explicitly not to analyze their reasons. Compa red to nonreasoners, reasoners predicted fewer winners of the games and pre dicted margins of victory that differed more from both the actual margins o f victory and the margins of victory predicted by experts. The relationship between expertise and reasons analysis, and the implications of the result s for other domains of prediction are discussed.