Predictor Representation and Prediction Strategies

An important difference between visual and numerical representation of the predictor is that the former entails a frame of reference against which the predictor can be evaluated, thereby facilitating reliance on the representativeness heuristic. The main hypothesis examined in Experiment 1 is that reliance on representativeness will increase when predictions are based on predictors that are represented visually rather than numerically. In addition, it is hypothesized that factors such as predictor validity, number of predictors, and amount of outcome feedback interact with predictor representation in determining the influence of representativeness on predictions. Experiment 2 examines the hypothesis that reliance on representativeness will increase if subjects are provided with a frame of reference for the predictor by informing them about the predictor′s scale. In both experiments, extremity and consistency of predictions are used as indicators for reliance on representativeness. The results of Experiment 1 indicate that: (1) Predictions are more extreme when they are based on visually represented predictors, but this effect depends on the amount of outcome feedback and disappears in high predictor validity, and (2) predictor representation interacts with the number of predictors in determining prediction consistency, but not in determining prediction extremity, but this interaction depends on predictor validity. Experiment 2 replicates Experiment 1 with regards to extremity but not with regard to consistency.