ON TESTING OF HYPOTHESES IN SEQUENTIAL ANALYSIS FOR A BINOMIAL DISTRIBUTION
Abstract
A binomial distribution with the parameter p is under consideration. The null hypothesis, Ho(p < po), will be tested against he lternative hypothesis H1(p > po) by the known sequential procedure which is broken off after No trials. The following conditions shall be satisfied: (1) H1 is accepted with a significance level < o for p < po, (2) Ho is accepted with a significance level < 2 for p > p2, (3) The error of the second kind 1 of the procedure is related to the corresponding p1 value by a function ( 1, p1) = 0 with po < p1 < 1, 0 < 1 < 1 - o, and (4) The hypothesis H1 is to be accepted, with a minimum of trials, uniformly for all p > po. The solution of this system leads to an extreme value problem, which may be solved under certain conditions. A special example is calculated for no = const with no as the maximum ASN, in which the error of the first kind = o for given po and this results in ( 1, p1) = 0. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Mar 15, 1962
- Accession Number
- AD0283455
Entities
People
- Eberhard H. Baur