Randomization Tests for Comparing Survival Curves

Publisher Website
Google Scholar
Abstract
Randomization tests are utilized to test the significance of differences in mortality observed between 2 groups of pigs, 1 sired by normal males, the other sired by males treated with X-rays (300 r). Random samples of experimental units are partitioned into groups comparable to the groups fromed in the actual experiment. The significance of differences found in the actual experiment is judged on the basis of the proportion of random partitions which yield equal or greater differences for the various criteria under study. The method is used because of the inherently complex correlational structure of the data that cannot be adequately described in simple parametric models. Results indicate a small and significant increase in early postnatal mortality among pigs sired by males exposed to X-rays compared with comparable controls.

This publication has 3 references indexed in Scilit: