Goal programming approaches for priorities in Saaty's analytic hierarchy process

A variety of alternative estimation procedures have been proposed, from the use of different means to the least square method and linear programming. The most common approach is the originally proposed eigenvector method, which is dominant in practical use, as numerous publications have demonstrated. Saaty compared the eigenvector method with the logarithmic least squares method and concluded with ten reasons for not using substitutes for the eigenvector, with ‘uniqueness’ of solution as first of them. The necessity for further research on estimation procedures is pointed out by Zahedi in his survey. Our paper proposes a goal programming approach different in its constructions from Korhonen and Wallenius' model as an alternative estimation procedure. By means of simulation results the approach's ability to produce unique solutions is demonstrated. As opposed to the other proposed methods, the goal programming approach offers the advantage of allowing a simultaneous estimation of all the decision elements under certain circumstances.