Most of the statistical procedures we’ve introduced in the previous 0 chapters require an assumption about the form of a probability distribution, often the Normal distribution. When such assumptions are unjustified, the consequences of the procedure are dubious at best. In situations where distributional assumptions cannot be justified, even after a well-chosen data transformation, the analyst should consider another approach.
KeywordsSign Test Whitney Test Distributional Assumption Population Median Parametric Procedure
- E. Lehmann, Nonparametrics—Statistical Methods Based on Ranks, revised first edition (Prentice Hall, New York, 1998)Google Scholar