This paper surveys the use of nonparametric permutation tests for analyzing experimental data. The permutation approach, which involves randomizing or permuting features of the observed data, is a flexible and convenient way to draw statistical inferences in many common settings. It is particularly valuable when few independent observations are available, as is often the case for controlled experiments in economics and other social sciences. When viewed as a framework, the permutation method constitutes a comprehensive approach to statistical inference. In two-treatment testing, permutation concepts underlie popular rank-based tests, like the Wilcoxon and Mann-Whitney tests. But permutation reasoning is not limited to ordinal contexts. Analogous tests are easily constructed for the permutation of continuous measurements, and we argue that these non-ranked alternatives should often be preferred when working with continuous data. Permutation tests can also be used with multiple treatments, with ordered hypothesized effects, and with complex data structures, such as hypothesis testing in the presence of nuisance variables. Drawing examples from the experimental literature, this paper illustrates how permutation testing solves common data analysis challenges. Our aim is to help experimenters move beyond the handful of overused tests in play today, and to show how permutation testing constitutes a general framework for conducting statistical inference with experimental data
