Title: There is Hope in Attribute Weighting
Authors: Mari Pöyhönen, Raimo P. Hämäläinen, Helsinki University of Technology
Date: March, 1998
Status: Systems Analysis Laboratory Research Reports A74

Experimental evidence shows that attribute weighting in value trees is prone to various biases. The origins of these biases are suggested to be either behavioral or procedural. It is most surprising that the literature does not have reports on biases observed in connection with applications. This is the situation both with multiattribute value tree analysis and the analytic hierarchy process. Can it really be true that biases are only experimental artifacts which do not occur in applications? We believe that the risks of biases exist but so far they have been ignored. Is there something a practitioner could do to avoid such risks? The answer is yes. We report how the increased understanding of the weight elicitation methodology and the awareness of the biases can decrease their occurrence.

Click here to access a pdf version of this paper.

Back to the Decision Analysis Working Paper Index