Generally Applicable Limited-Length Sequential Permutation Tests for One-Way Anova.
Abstract
The data are independent observations that, under the null hypothesis, have the same unknown distribution (which can be arbitrary). Observations are obtained in sets of specified sizes, with a maximum total number available. An overall test is conducted as a succession of subtests and is significant when at least one subtest is independent of previous subtest results. The overall test is terminated when a significant subtest occurs (thus saving time and expense). Subtests are always included wherein the second and each following set is compared with the totality of previous observations to investigate whether they continue to be from the same population. If desired, an additional subtest can be included so that the first set is investigated for the random sample hypothesis. The alternative hypotheses emphasized are determined by the subtest statistics. Unconditional tests can be obtained when ranks are used. There are indications that, by suitable choice of the subtest statistics, these tests compare favorably with tests using the maximum number of observations. One application is for quality control and some possible uses are outlined. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Sep 10, 1970
- Accession Number
- AD0715085
Entities
People
- John E. Walsh
Organizations
- Southern Methodist University