SPC Guide
Ask Experts
First Word


Michael J. Cleary, Ph.D.

Nothing Succeeds Like Excess

Even Oscar Wilde had but a limited understanding of sampling.

Simply driving past a manufacturing plant or seeing a photo of an assembly facility, few people are aware of the seething conflicts and deep-seated passions that lie within. The gentle fašade of a modern brick facility belies the frenetic pace and the heated statistical arguments that ensue among those who work in this seemingly serene setting. As we know from our observation of Greer Grate and Gate and its quality team, the assumption that all is placid within couldn't be further from the truth.

 Hartford Simsack, Greer Grate and Gate's intrepid quality manager, is glowing with self importance after a succession of successful applications of statistical knowledge. You will recall that he has been right twice--first, in instructing quality technician Franklin Benjamin about control limits, and then in rescuing a supplier, Rhott Iern Productions, by teaching quality manager Ms. X. Trusion about using chi-square tests to determine process normalcy.

 Simsack's little secret--that he relies on his statistics professor at the community college to give him the right answers to questions that arise--has not been discovered, and he is garnering something of a reputation as a statistical expert. He polishes his knowledge by reading the daily statistical information offered in USA Today. Or, at least, he looks at the pictures.

 In the meantime, our friend Benjamin has been reading statistics books during his lunch breaks, and his questions for Simsack are becoming increasingly sophisticated. Only yesterday, Hartford had to fake losing his voice in order to delay providing an explanation of Pearson curve fitting.

 Benjamin would like to increase the amount of data collected from his production line, believing that having more data will improve accuracy. He's from a large family and has grown up with a "the more the merrier" approach to most things--or as Oscar Wilde put it, "Nothing succeeds like excess." Benjamin believes that checking every second piece that emerges from the production line would do it. (He would prefer to check every piece, but he believes that a compromise is in order.) Data is currently gathered from every 10th piece. "We can increase reporting accuracy by five times the current rate," he tells Simsack, who nods solemnly and puffs on his pipe in the same way he has seen his professor and mentor, Dr. Stan Deviation, do. He makes a mental note to ask Deviation about the sampling and to find out what kind of pipe tobacco he uses.

 Because the measurement process for GG&G's products doesn't involve destructive testing (the way it might in a wooden match factory, for example), Simsack thinks it can do no harm to increase the sampling rate. He agrees with Benjamin that generating more charts would improve the process. Secretly, he thinks this move will make him look good as well. Simsack remembers seeing photos of Japanese plants that had created hundreds of control charts and hung them on the walls, long before he even knew the charts' purpose.

 When Simsack shared Benjamin's enthusiasm for increased frequency of sampling with Deviation, the professor laughed. "That guy doesn't have a clue," he chuckled. Simsack laughed nervously, trying not to reveal that he had already approved Benjamin's plan. What is the problem with stepped-up sampling in the way that Benjamin wants to pursue it?



 Sampling guides the quantitative study of a system and is a critical way to evaluate quality. A subset of a population, if it's approached correctly, gives the same information as might be found in an entire population, with less investment in time and money. Appropriate sampling frequency is critical to its accuracy. No single determination of sampling frequency is right for every process and every chart.

 To get the best information about a critical factor, first determine what one needs to know from the data. Then two questions must be answered:

1. How often do things change? The more often a process changes, the more frequently sampling should be done. If there are cycles in the data, samples should be taken both during and between peaks and valleys. Consider changes of shift, equipment and materials. Frequency can be stated in time (hourly, daily, etc.) or in number (e.g., every 10th piece).

2. How much does sampling cost? If extensive resources are required, this must be a factor in determining intervals. The benefit of additional information is weighed against the cost of collecting the data.


 "More is better" does not make sense at Greer Grate and Gate. The relevant factors are the frequency of process changes and the costs of sampling. Process operators already know this, Deviation knows it, and (once it's pointed out to him) Simsack knows it.

 The question is, how is Simsack going to break the news to Benjamin?



 A classic reference for frequency (and size) of sampling is Statistical Quality Control Handbook (Western Electric Co., 1956).


About the author

 Michael J. Cleary, Ph.D., is founder and president of PQ Systems Inc. He has published articles on quality management and statistical process control in a variety of academic and professional journals. E-mail Cleary at .

Menu Level Above 

This Menu LeveL 

Menu  Level Below 

[Contents] [News] [WebLinks] [Columnists]
[Applications] [Software] [SPC Guide] [Letters] [Ask Experts] [First Word] [Books]

Copyright 2000 QCI International. All rights reserved.
Quality Digest can be reached by phone at (530) 893-4095. E-mail:
Click Here

Today's Specials