Testing Decision Rules for Multiattribute Decision Making

This paper investigates the existence of an editing phase and studies the com- pliance of subjects' behaviour with the most popular multiattribute decision rules. We observed that our data comply well with the existence of an editing phase, at least if we allow for a natural error rate of some 25%. We also found a satis- factory performance of certain groups of subjects for the conjunctive rule, for the elimination{by{aspects rule, for the majority rule, and for the maximin rule. Our data suggest, however, rejection of the prominence hypothesis and of the maximax rule. Thus, our experiment sheds light on the existence of an editing phase and on the use of various multiattribute decision rules.