An investigation of sensitivity analysis as an analytical tool

The purpose of this study was to examine sensitivity analysis as an analytical tool. The method of procedure was to examine the definition, use and techniques of sensitivity analysis. In addition the application of sensitivity analysis was illustrated on models tractable to partial derivative, simulation and linear programming solution methods. As opposed to an analysis of specific problems, the study was a procedure oriented approach. Sensitivity analysis is useful in identifying variables which require precise estimates, and in observing the best and worst possible outcomes of decisions under uncertainty. Three steps were identified which constitute an analysis: (1) determine which variables to examine; (2) establish the range of each variable examined; and (3) determine the measure of sensitivity. The problems of size of the model and interaction of variables are limiting factors in an analysis. Experimental design and linear programming range techniques were illustrated as procedures for resolving such limitations.