A Belief-Updating Perspective on Advice Taking

Advice taking is the typical mode of social information sampling. We will examine advice taking in a prototypical price estimation scenario, where advice in form of price estimations from previous participants can be sampled in order to update a pre-advice estimate to a final price estimate. Numerical advice-taking literature was strongly oriented towards simple averaging as a benchmark model. In this study we take a (Bayesian) belief-updating perspective on the informed guessing task. The belief-updating perspective implies that discounting distant advice is not per se an indicator of egocentric bias. Rather, whether or not advice is followed or not depends on how consistent the sampled advice appears. This leads to the interplay of two crucial characteristics of advice: distant (vs. close) advice and homogenous (vs. heterogeneous) advice cause stronger impression change, higher levels of confidence in the final estimate and earlier stopping in seeking advice.