Sequential Multivariate Quality Control Tests Using Tolerance Regions
SOUTHERN METHODIST UNIV DALLAS TX DEPT OF STATISTICAL SCIENCE
Pagination or Media Count:
Two new developments, multivariate sequential significance tests and a method of forming multivariate two-sample tolerance tests, are proposed. The sequential significance test considered is a fixed-length succession of two-sample subtests where each subtest reuses some or all of the data for preceding subtests. By proper choice of the subtest statistics and use of a permutation basis the subtests are made independent. The class of multivariate two-sample tolerance tests developed by the proposed method are directly applicable as subtests in the sequential significance tests. The proposed method is based on a new technique of constructing tolerance regions for the two-sample problem. Subject to certain mild limitations the analyst may actually look at the combined observed vector values in order to construct the desired tolerance regions.
- Statistics and Probability