Goal programming and cognitive biases in decision-making

The study of cognitive biases in decision-making has largely arisen within the context of the subjective expected utility school of decision analysis. Many of the behavioural patterns that have been discovered do seem to be relevant to broader areas of multicriteria decision analysis (MCDA). In this paper, we look specifically at the judgemental inputs required in implementing goal programming (GP) models. The potential relevance of some of the known cognitive biases in this context are identified, and their impact studied by means of simulation experiments. It is found, inter alia, that biases due to anchoring and adjustment and to avoidance of sure loss can lead to substantial degradation in the performance of GP algorithms. Suggestions for practice and recommendations for follow-up research are derived from the simulation results.