Cpk definition for EBM and SBM bottles,

[mcpe72]

they know its i think is more a technicality; can we just tell them that because its n 30 samples we assume normality because of the central limit theorem?

 

[Bev D]

No as that is not true. At all.
The central limit theorem applies to the distribution of the averages of many sample sets (formally known as subgroups) there are typically about 20 or so subgroups for the central limit theorem to be roughly in play.

And how does assuming normality help you?

 

[mcpe72]

this is what i found,

Link to free article: “The Assumption(s) of Normality” by J. Toby Mordkoff, Copyright © 2000, 2011, 2016

 

[mcpe72]

The long and formal answer to this question relies on Central Limit Theorem which says that: given random and independent samples of N observations each, the distribution of sample means approaches normality as the size of N increases, regardless of the shape of the population distribution. Note that the last part of this statement removes any conditions on the shape of population distribution from which the samples are taken. No matter what distribution you start with (i.e., no matter what the shape of the population), the distribution of sample means becomes normal as the size of the samples increases. (I’ve also seen this called “the Normal Law.”)

 

[Bev D]

Yeah, so? It just says what I already said. You do not have enough subgroups of n>30 each to apply this. If you sampled from multiple cavities you don’t have ‘random’ (or sampling from the same population of values) because there is always a systemic difference between cavities. And more importantly Cpk isn’t about sample averages but individual values. So the question still stands: how could this possibly help you?

 

[mcpe72]

because we are in the phase of development and qualification, i can use the 30 samples and don't group them

 

[Bev D]

Huh? That makes no sense. The Central limit theorem doesn’t ’allow’ you to do this, nor does it justify it. It would be helpful if you could explain why you think the central theorem justifies this…most of us here know what the central limit theorem is but I for one have no idea why you think it is helpful…

What do you mean by grouping? How did you structure your original 30 parts in minitab? (Your Cpk and Ppk values don’t make sense)
I could guess what you did in ‘grouping’ but It’s better if you tell us. I also think you and/or your Customer don’t understand what is needed for a development process. It is Ppk* (Ppk formula used as short term “capability”. We have discussed this paradox many times here and if needed one of us can provide a link to those discussions)

And none of this is effected at all by the central limit theorem.

* although Cpk and Ppk are practically useless and if anything they are simply misleading. But that not withstanding, many people have hitched their ‘quality’ wagon to this statistical alchemy…and those that have usually don’t understand it.

 

[Ed Panek]

You cant have it both ways. You cant be "We are in the sandbox mode of development," And "'But I want these sandbox tests to count"

 

[Alex_beasley]

Just to let everyone know this thread was very helpful to me in my current situation. Learned a lot by reading through all the comments.

As a seasonal professional in quality and manufacturing i will like to add that capability studies does not only provide product stability but it also helps in reducing your COPQ, it derives the organization to be customer focused. So it's not about the criticality of product but also about where the organization wants to see it self .