THE ANALYSIS OF VARIANCE IN CASES OF NON-NORMAL VARIATION
- 1 November 1931
- journal article
- research article
- Published by Oxford University Press (OUP) in Biometrika
- Vol. 23 (1-2) , 114-133
- https://doi.org/10.1093/biomet/23.1-2.114
Abstract
The results of experimental sampling suggest that the customary analysis of variance or test of the significance of n2 is adequate for a wide range of non-normal populations. In applying the test we shall not be liable to err by rejecting the hypothesis of a common origin of the groups when it is true. But in the more extreme cases of non-normal variation, there is danger of accepting the hypothesis when it is not true. The case in which the variation changes from group to group is not considered. The test of significance of the differ ence in the variances of 2 independent samples is much more sensitive in small samples to changes in population form than is the n2 test. The more elaborate forms of analysis of variance are probably also of fairly wide application, if the number of degrees of freedom apportioned to the residual variation is not too small.Keywords
This publication has 0 references indexed in Scilit: