The devil is in the specificity: the negative effect of prediction specificity on prediction accuracy.

In the research reported here, we proposed and demonstrated the prediction-specificity effect, which states that people's prediction of the general outcome of an event (e.g., the winner of a soccer match) is less accurate when the prediction question is framed in a more specific manner (e.g., guessing the score) rather than in a less specific manner (e.g., guessing the winner). We demonstrated this effect by examining people's predictions on actual sports games both in field and laboratory studies. In Study 1, the analysis of 19 billion bets from a commercial sports-betting business provided evidence for the effect of prediction specificity. This effect was replicated in three controlled laboratory studies, in which participants predicted the outcomes of a series of soccer matches. Furthermore, the negative effect of prediction specificity was mediated by participants' underweighting of important holistic information during decision making.