Are Non-Parametric Tests Distribution-Free,
Abstract
An increasing use is being made of non-parametric tests, i.e., tests of significance which are not based on the usual assumption that all distributions are normal. Many psychologists have assumed that when non-normality or heterogenous variance is encountered in the data, then analysis of variance by ranks, the Mann-Whitney U Test (1947), Festinger's d (1946), or some similar non-parametric technique is appropriate. The purpose of this paper is to point out that non-parametric tests also have basic assumptions which must be met, and that often the factors which bar the use of the parametric tests also violate the assumptions underlying most of the non-parametric tests.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jan 01, 1961
- Accession Number
- ADA074099
Entities
People
- Ardie Lubin
Organizations
- Walter Reed Army Institute of Research