Sampling Size and Sampling Frequency

[Semoi]

@hazwan2283: In this forum you will find plenty of references to Donald Wheelers articles. Reading his articles you will often find examples where the process is in the "developing phase" and therefore inhomogeneous. If you are in this phase, following Wheelers recommendations is a good idea. However, if you possess an optimised process and your main goal is to implement a process monitoring, I recommend reading Montgomery's books. He is best known for his "Design of Experiment", but he also wrote books about statistical process control. I know that he recommends transformations.

PS: I also know that Wheeler does not agree with the two phases: developing and monitoring.

 

[Bev D]

Stable and capable processes can be non-homogenous. Hence teh use of rational subgrouping.
A homogenous process is one where sequential piece to piece variation is related to teh sub group to subgroup variation: SD_Average = total SD/sqrt. In other words the factors that control the average also control the standard deviation. Understanding the components of variation is essential to understanding how to subgroup, the frequency of sampling, and the sample size(s). See this article: The Three-Way Chart. Leonard Seder also describes this very well in the context of problem solving. His articles are not free but can be found and purchased via a simple internet search.

There are some who will say that an ideal process is one that is homogenous. But today’s processes and products are quite complicated and often naturally non-homogenous. (For example, within batch variation is not predictive of batch to batch variation in most chemical or biological processes). Of course we can work to reduce the variation that lies above piece to piece but this isn’t always economically warranted if the non-homogenous process is stable and capable…