A sensitivity analysis algorithm for the constant sum pair-wise comparison judgments in hierarchical decision models

Every hierarchical decision modeling process starts with quantifying the contributions of decision elements through pair-wise comparisons. As subjective values, the pair-wise comparison judgments are seldom provided at a 100% confidence level and are subject to variations. To increase the model´s validity and ensure requisite decision making, it is important to know how sensitive the model result is to these inputs. In this paper, a sensitivity analysis algorithm is developed to test a hierarchical decision model´s robustness to the pair-wise comparison judgment inputs acquire from the constant sum method. It defines the allowable region of perturbation(s) induced to a judgment matrix at any level of a decision hierarchy to keep the current ranking of decision alternatives unchanged. An example will be presented to demonstrate the application of this algorithm in technology selection.