Influence of group members in multi-attribute utilities

摘要

This paper investigates measuring the influence of some group members on others in decision making. Being better able to identify potentially influential behaviour would be useful in supporting and subsequently auditing a decision. A new measure of the influence of individuals is given, which is analogous to the well-known Cook's distance used to identify influential data in regression. The theoretical properties of this measure are explored. A simple method to identify sub-groups within the group of decision makers is given. We investigate the efficiency of our new measures using large scale randomised studies. We use these measures to identify sub-groups of individuals with similar beliefs in a data set collected in a previous experiment.