It is the analysis of the impact of limiting assumptions made in a model on the results of an analysis.