Document Type

Thesis - University Access Only

Award Date


Degree Name

Master of Science (MS)

Department / School

Industrial Management

First Advisor

Robert J. Lacher


In this era of intense competition, understanding and improving quality is of paramount import toward achieving business success and enhanced competitive edge. A process capability index (PCI) has been considered to be a valuable tool to express the capability of processes and assure the quality of products. In recent times process capability indices have received substantial attention in statistical and quality control publications. The focus of the research work is to study various process capability indices - Cp, Cpk, Cpc, and their relative advantages and limitations. It tests Cpc' s reported claim of its superiority over other indices concerning their sensitivity to departures from the assumption of normal distribution. The simulation was performed with the help of a software program, Resampling Stats(R), through generation of simulation runs. In this respect, two population distributions were considered, i.e., normal and exponential. Two values of sample size (n = 5 and n =125) were taken to ascertain sample-size-effect on the two distributions. Because the standard deviation is the same as the m􀄡an for exponential distributions, a change in the value of standard deviation, ( cr = 10 and cr = 35) was considered in case of normal distribution only. From the results, it was found that unlike Cpk, which fit on well with normal distribution, but do􀄢s not in case of exponential distribution, Cpc, the third generation index, is more consistent with exponential distribution than normal distributions. This was true for a small sample size, n = 5. With a large sample size, n = 125, Cpc seemed to perform better for normal than exponential distributions. So, for a bigger sample size, Cpc acted like Cpk• As Cpc changed its nature of behavior vis-a-vis distributions when sample size changes, it appeared to contradict third generation index's superiority over the first one concerning the indices' sensitivity to departures from the assumption of normal distribution. The effect of the change in the value of standard deviation ( a = 10 and a = 3 5) for normal distribution indicated that Cpc fit better with a higher value than a lower one. It is hoped that this research work will be helpful in further examining the nature of various process capability indices.

Library of Congress Subject Headings

Quality control -- Statistical methods.
Process control -- Testing.


South Dakota State University



Rights Statement

In Copyright