ATTRIBUTE WEIGHTING METHODS AND DECISION QUALITY IN THE PRESENCE OF RESPONSE ERROR - A SIMULATION STUDY

This paper uses a simulation approach to investigate how different att ribute weighting techniques affect the quality of decisions based on m ultiattribute value models. The weighting methods considered include e qual weighting of all attributes, two methods for using judgments abou t the rank ordering of weights, and a method for using judgments about the ratios of weights. The question addressed is: How well does each method perform when based on judgments of attribute weights that are u nbiased but subject to random error? To address this question, we empl oy simulation methods. The simulation results indicate that ratio weig hts were either better than rank order weights (when error in the rati o weights was small or moderate) or tied with them (when error was lar ge). Both ratio weights and rank order weights were substantially supe rior to the equal weights method in all cases studied. Our findings su ggest that it will usually be worth the extra time and effort required to assess ratio weights. In cases where the extra time or effort requ ired is too great, rank order weights will usually give a good approxi mation to the true weights. Comparisons of the two rank-order weightin g methods favored the rank-order-centroid method over the rank-sum met hod. (C) 1998 John Wiley & Sons, Ltd.