ABSTRACT
Sensitivity analysis is the study of how the variation in a risk assessment output can be apportioned, qualitatively or quantitatively, to different sources. Qualitative sensitivity analysis begins by identifying specific sources of uncertainty then ascertaining the sources of instrumental uncertainty. A WHO scheme for conducting a qualitative sensitivity analysis is described. Three approaches to quantitative sensitivity analysis are described. Mathematical methods described include nominal range sensitivity, difference in log-odds ratios, break-even analysis, and the automatic differentiation technique. Statistical methods including regression and correlation analysis, ANOVA, and others are described. Scatter, tornado, and spider plots are among the graphical methods for sensitivity analysis described. A primary purpose of sensitivity analysis is to inform strategies for reducing instrumental uncertainty.
