rbtt

Tu & Zhou (1999) doi:10.1002/(SICI)1097-0258(19991030)18:20%3C2749::AID-SIM195%3E3.0.CO;2-C showed that comparing the means of populations whose data-generating distributions are non-negative with excess zero observations is a problem of great importance in the analysis of medical cost data. In the same study, Tu & Zhou discuss that it can be difficult to control type-I error rates of general-purpose statistical tests for comparing the means of these particular data sets. This package allows users to perform a modified bootstrap-based t-test that aims to better control type-I error rates in these situations.