![]() ![]() Sample sizes can also be small in natural experiments, especially where there is little exogenous dispersion of treatment conditions. These costs were magnified during the recent Covid-19 lockdowns that required interactive experiments to be run online, with a significant fraction of Zoom sessions being interrupted by subjects leaving the meeting or experiencing connectivity issues. The costs of recruiting and incentivizing subjects to participate in research experiments often force experimenters to settle for fewer observations than we might like. The second challenge is small sample sizes. Markets and other group interactions create dependence relationships between observational units. In economics and other social sciences, data from laboratory and field experiments present two common challenges for statistical inference. Our aim is to help experimenters move beyond the handful of overused tests in play today and to instead see permutation testing as a flexible framework for statistical inference. Drawing examples from the experimental economics literature, we illustrate how permutation testing solves common challenges. 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. Analogous tests can be constructed from the permutation of measured observations-as opposed to rank-transformed observations-and we argue that these tests should often be preferred. But permutation reasoning is not limited to ordinal contexts. In two-treatment testing, permutation concepts underlie popular rank-based tests, like the Wilcoxon and Mann–Whitney tests. The permutation method constitutes a comprehensive approach to statistical inference. It is particularly valuable when few independent observations are available, a frequent occurrence in controlled experiments in economics and other social sciences. The permutation approach, which involves randomizing or permuting features of the observed data, is a flexible way to draw statistical inferences in common experimental settings. This article surveys the use of nonparametric permutation tests for analyzing experimental data. ![]()
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |