SINGLE-STAGE PROCEDURES FOR RANKING MULTIPLY-CLASSIFIED VARIANCES OF NORMAL POPULATIONS.
CORNELL UNIV ITHACA N Y DEPT OF OPERATIONS RESEARCH
Pagination or Media Count:
The author and Sobel have proposed single-stage procedures for ranking the variances of normal populations which are classified according to a single qualitative factor. Using the multiplicative model for variances introduced by the author in an earlier paper, it is shown how to generalize these procedures so that they can deal with situations in which the variances are classified according to two or more qualitative factors. The rationale of such experiments is discussed. Goals and associated probability requirements and procedures are described. Explicit expressions for the exact probability of a correct selection, using the proposed procedures, are obtained for certain special cases. The determination of sample size to guarantee the probability requirement is considered, large-sample approximations to these sample sizes are given, and the accuracy of the approximations is assessed. Author
- Statistics and Probability
- Operations Research