From: Stan Hilliard
Date: 08 Jun 2000
Time: 01:49:10
Greetings Larry,
You said: "Since "N" is drawn at random from a theoretical infinite process and "n" is a random sample of "N" then you have in fact a random sample "n" from the process."
But to say that is to assume that the non-theoretical real process does not have off-grade sections related to natural sublots.
A related issue: When a process is said to be in control, that means that it was in control in the past -- during the time period that the data was collected. It does not mean that it will remain that way.
Acceptance sampling plans are application of hypothesis tests. In hypothesis testing, you don't hypothesize that the situation that you want to prove is true. You hypothesize that the situation that you want to disprove is true. This puts the burden of proof on the data demonstrate with confidence what you want prove.
This is identical logic to a court of law. The prosecution wants to prove that the accused guilty of the crime, but the law says that the null hypothesis is innocence and puts the burden of proof on the evidence (data) to demonstrate otherwise. This appears to many to be twisted form of logic, but it is correct and very powerful.
Sincerely, Stan Hilliard