Performance of the AHP in comparison of gains and losses

In this paper, we have empirically investigated performance of the analytic hierarchy process (AHP), when the utility of the objects cannot be evaluated on the same ratio scale. The basic assumption in the AHP is that a decision maker compares objects on a ratio utility scale. However, there are situations in which the AHP is (accidentally) used even if the basic assumption is violated. This kind of the problem occurs, when a decision maker is asked to compare, for instance, the objects (s)he likes to the objects (s)he hates. In this case, the performance of the AHP is expected to be very poor, because each object the decision maker likes is presumably “absolutely better” than any object (s)he hates. Our original purpose was to demonstrate empirically that comparing gains and losses ruins the AHP-analysis. However, our results demonstrate that the AHP is able to estimate the reasonable utility values for objects surprisingly well. The origin separating utility and disutility scales was estimated as well.