Promoting Intersubjectivity: A Recursive-Betting Model of Evaluative Judgments

Evaluative judgement aims at intersubjectivity, rather than objectivity or subjectivity. Multiple judges are widely used to increase its common component. A practical question is how to aggregate the individual judgements, in view of the feed-forward effects that are elicited by aggregation policies. The traditional practice of taking the mean of the individual judges’ votes, ratings, or rankings invites improper strategic manoeuvring by judges. The development of a set of more proper aggregation rules requires an analysis of the judgement task, as distinct from measurement, forecasting, preferential choice, and voting. Using psychometric notions, I model judgement as a recursive bet aimed at maximising the common or intersubjective component in the individual judgements. For practical purposes, the analysis leads to a number of recommendations, in particular: Take the median, not the mean, in aggregating individual evaluative judgements, and inform judges about its rationale. (Netherlands Journal of Psychology 65, 2-9.)