Eliciting Human Judgment for Prediction Algorithms

Even when human point forecasts are less accurate than data-based algorithm predictions, they can still help boost performance by being used as inputs. Assuming one uses judgment indirectly in this manner, we propose changing the elicitation question from traditional direct forecast (DF) to what call private information adjustment (PIA): how much thinks should adjust its account for has that is unused algorithm. Using stylized models with and without random error, theoretically prove error makes eliciting PIA lead more predictions DF. However, DF-PIA gap does not exist perfectly consistent forecasters. The increasing people make while incorporating public (data uses) but decreasing only use). In controlled experiments students Amazon Mechanical Turk workers, find support these hypotheses. This paper was accepted Charles Corbett, operations management.