Wilcoxon Signed-Rank Test in Statistical Literature and Practical Use
View Presentation View Presentation
Terri Johnson, The US FDA  *Heng Li, FDA/CDRH 

Keywords: non-parametric

Since it was proposed in 1945, the Wilcoxon signed-rank test (or Wilcoxon one-sample test) has become one of the most recognized statistical techniques in hypothesis testing. However, it is not always dealt with sufficiently thoroughly in introductory or applied statistical texts. While steps in obtaining the test statistic are provided in detail and illustrated with numerical examples, the statement of hypotheses and assumptions can be relatively vague. There have also been experiences in practice where intended usage of the test was somewhat incorrect. In this presentation we will conduct an in-depth discussion on the proper use of Wilcoxon signed-rank test. We will illustrate the behavior of the test when its assumptions are not met, and compare its performance to other statistical tests.