We have David Budescu, Fordham University, USA, visiting and giving a talk in the Social, Economic, and Decision Psychology Colloquium (talk and abstract follow).

Identifying Expertise and the Wisdom of Selected Crowds

The term “wisdom of crowds” (Surowiecki, 2004) is often used to describe the robust empirical finding that statistical aggregates of the opinions or estimates of the group’s members (i.e., methods that do not involve direct interactions among the members) outperform most individual judgments. In the first part of this talk I will offer a general definition of the “wisdom of the crowd effect”’ as well as a statistical framework in which to evaluate it, that explicitly accounts for the inter-dependency among the members of the crowd and their biases. Crowd prediction is treated as a linear combination of group member prediction distributions and the average performance of this aggregate prediction is compared to an individual member (or group of members) selected according to an arbitrary, pre-specified probability distribution. In the second part of the talk I will discuss new methods to measure the contributions of the various individuals to the crowd. I propose using a variant of the influence function that differentiates between the individual contributions, identifies the best (and the worst) contributors and allows one to derive differential weights for the various group members. These measures of individual contributions can be used to derive improved aggregation schemes that outperform the regular averaging procedures. The procedure is illustrated and validated with date from a large–scale project involving forecasts of geopolitical events.