We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
The Analysis of a 2n Experiment by Means of Ranks.
- Authors
MacDonald, Peter
- Abstract
SUMMARY A method is proposed for the analysis of variance of ranked data from a factorial experiment of size 2[sup n] which may be replicated or not. A test of the hypothesis that a given linear contrast of means is zero is based on the corresponding contrast of the ranks of the observations in each replicate. An approximate critical region for the test is based on the rank sum test, which is equivalent to the new test under certain null hypotheses. Two other forms of test are also suggested. Estimates of the population contrasts may be obtained by means of an ad hoc procedure for resolving the ambiguity in the mean of the test statistic, together with normalization.
- Subjects
ANALYSIS of variance; RANKING (Statistics); FACTORIAL experiment designs
- Publication
Journal of the Royal Statistical Society: Series C (Applied Statistics), 1971, Vol 20, Issue 3, p259
- ISSN
0035-9254
- Publication type
Article
- DOI
10.2307/2346756