The design of a poverty measure involves the selection of a set of parameters and poverty figures. I in most cases the measures are estimated from sample surveys. This raises the question of how conclusive particular poverty comparisons are subject to both the set of selected parameters (or variations within a plausible range) and the sample datasets. This chapter shows how to apply dominance and rank robustness tests to assess comparisons as poverty cutoffs and other parameters changes. It presents ingredients of statistical inference, including standard errors, confidence intervals, and hypothesis tests. And it discusses how robustness and statistical inference tools can be used together to assert concrete policy conclusions. An appendix presents methods for computing standard errors, including the bootstrapped standard errors.