NCJ Number
102564
Date Published
1982
Length
32 pages
Annotation
This paper presents a model for evaluating decision alternatives characterized by intermediate levels of multidimensional attributes.
Abstract
Complex decision problems are often modeled using alternative factorial representations of the decision alternatives. This is frequently accomplished by using multiattribute value models of the types described in Keeney and Raiffa (1976) and Dyer and Sarin (1979). Under these models, the attributes of alternative decisions must have a degree of preferential independence. When these conditions are not met, it is sometimes possible to redefine a set of attributes to constitute multidimensional attributes. This raises the problem of consistently characterizing intermediate levels of multidimensional attributes which are individually comprised of several dependent unidimensional attributes. This paper describes a procedure for developing an assessment data base using a limited number of observations of a decisionmaker's (DM's) preference space. This data base is then used as a basis for using interpolation models to evaluate intermediate multidimensional attribute levels. The primary advantage of the procedure is its capacity to estimate higher order preference dependencies based on a limited number of observations of the DM's preference space. Mathematical equations, figures, and 4 references. (Author abstract modified)