Experimental test of MCDM algorithms in a modular approach

The Stem algorithm and Steuer's filtering procedure were offered to 142 subjects to assist them in solving six car buying problems of various complexity. Subjects could change between the algorithms and some other program options almost at will. We show that the algorithms are evaluated differently and that objective measures of the solution process show differences between the algorithms that support the subjective evaluations. If the objective task complexity is noticed (at least along one dimension: number of objective functions) we find that the choice of procedure varies with it as well as with subjective complexity evaluations. In general, Stem is favored by a minority of subjects, preferably for smaller sized problems.