Approximate measurement in a multiattribute utility context

Abstract

W. G. Stillwell, D. A. Seaver, and W. Edwards (Organizational Performance and Human Behavior, 1981, 28, 62–77) found, among other things, that a simple ranking method for eliciting weights in multiattribute utility (MAU) contexts produced orderings that were good approximations to those produced by a more rigorous, but more complicated ratio method. Because simple methods are desirable for applied work, the present research examined the degree to which three other simple methods yield good approximations to the orderings produced by the aforementioned ranking method. At the behest of a group charged with the long-term planning of a large county's park development, we had developed a MAU-based preference questionnaire which was used both in a survey of county residents' preferences and in a laboratory study; results of both of which are presented. It was found that the three simple methods—(1) point allocation, (2) voting, and (3) rating—all yielded preference orderings that were fairly good approximations to the orderings yielded by the ranking method, but that the voting method appeared to be the least satisfactory.