Olga Yermakhanova

Multicriteria Decision Support Systems simplify the process of decision-making in situations involving large numbers of criteria, constraints, alternatives, and colliding interests. Although evaluation methods vary, the output of a multicriteria analysis process is a set of evaluation scores, received by aggregation of standardized and weighted criterion values. Multicriteria methods are often used in the spatial context without understanding of how the spatiality in criterion values affects the spatial distribution of evaluation scores and sensitive alternatives.

This research studied the dependency of the autocorrelation and spatial distribution of the scores and sensitivities on the autocorrelation and intercorrelation of model criteria. A hypothetical multi-criteria decision problem, concerned with selection of locations for the implementation of new healthcare programs was employed to illustrate this.

This study found that autocorrelation of alternative scores and alternative sensitivities were best explained by the relationship among the criteria. The degree of autocorrelation of the scores was determined by intercorrelations among the criteria and by criterion weights. The clustering of alternative sensitivities and the number of sensitive alternatives showed little association with the autocorrelation of criteria. They depended on the size and prevalent type of sensitivity clusters, which in turn depended upon the cumulative distribution of criterion values, particularly on the size and type of overlapping criterion clusters.