CONDITIONAL RANK-ORDER TESTS FOR EXPERIMENTAL DESIGNS.
Abstract
The present work investigates the asymptotic efficiency of certain rank-order tests for testing the absence of main-effects in linear models under additivity. A distinct feature of these tests, unlike some previously available rank-tests of Friedman, Durbin, etc., is that they are based on a joint-ranking of the totality of observations after 'alignment', and yet can be applied conditionally (given epsilon = the ordered set of ranks for each block) at exact significance level for all sample sizes.
Document Details
- Document Type
- Technical Report
- Publication Date
- Feb 20, 1968
- Accession Number
- AD0666100
Entities
People
- K. L. Mehra
Organizations
- Stanford University