Min Tang


In evaluation of the operating characteristics of mixed sampling plans (a known), the probability function, Fn(x) , of the difference between the extreme value and the mean is important. Because it is an n multiple integral where n is the first sample size of a mixed dependent sampling plan, the normal way of evaluating this probability is the recursive quadrature method. Using this method, as the first sample size increases, the amount of computation increases exponentially, so that this probability function becomes a bottleneck in computation. This thesis presents a Taylor-expansion method for handling the probability function for large first sample sizes with satisfactory precision, for an arbitrary given x. In such a way, the evaluation of the dependent mixed plans is extended to a larger first sample size than previously available in the literature. Theoretically, the computational method presented here could be used for evaluating mixed plans of very large first sample size. Technically, however, this method is limited, by the computer memory resources. Using the computational approach described above, a FORTRAN program that can compute the operating characteristics of mixed plans for first sample sizes of up to 50 was developed. A series of mixed dependent sampling plans were then evaluated with an accuracy of seven decimal places.

Library of Congress Subject Headings

Sampling (Statistics)

Publication Date


Document Type


Department, Program, or Center

School of Mathematical Sciences (COS)


Schilley, Edward

Advisor/Committee Member

Lawrence, Daniel

Advisor/Committee Member

Wiet, Thomas


Note: imported from RIT’s Digital Media Library running on DSpace to RIT Scholar Works. Physical copy available through RIT's The Wallace Library at: QA276.6T35 1991


RIT – Main Campus