Determining Cpk based on a given Sample Size, Confidence Level, & Reliability

L

Lasermike

Hello, I'm new here and I'm hoping someone can help me out or point me in the right direction because I've spent some time reviewing this website and haven't found a clear answer to my question.

What is the way to determine a required Cpk based on a given sample size, confidence level, and reliability? At my last company we used this method to determine accecptance criteria for verification and validation activities and had sampling plan tables to guide the selection based on this critieria. For example, if I was trying to demonstrate that a process or product met a desired measureable output with 95% confidence/ 99.7% reliability, what would be the minimum Cp and Cpk (for two-sided spec) required for a sample size of 15 to demonstrate that level of confidence/reliability? We had tables and tables of different sample sizes and what Cps and Cpks would be needed to demonstrate different reliability and different confidence levels but I never fully understood the fundamental equations that drove the creation of the tables.

Thanks for anyone's help in advance!
 

Jim Wynne

Leader
Admin
Hello, I'm new here and I'm hoping someone can help me out or point me in the right direction because I've spent some time reviewing this website and haven't found a clear answer to my question.

What is the way to determine a required Cpk based on a given sample size, confidence level, and reliability? At my last company we used this method to determine accecptance criteria for verification and validation activities and had sampling plan tables to guide the selection based on this critieria. For example, if I was trying to demonstrate that a process or product met a desired measureable output with 95% confidence/ 99.7% reliability, what would be the minimum Cp and Cpk (for two-sided spec) required for a sample size of 15 to demonstrate that level of confidence/reliability? We had tables and tables of different sample sizes and what Cps and Cpks would be needed to demonstrate different reliability and different confidence levels but I never fully understood the fundamental equations that drove the creation of the tables.

Thanks for anyone's help in advance!

Welcome to the Cove. :D

While I'm sure there are formulae to get you where you want to be, trying to back-calculate a significant value by way of a dubious statistic like Cpk is probably not going to helpful in the long run. Unless the basis of the Cpk calculation conforms to the basic assumptions (a statistically stable process, chronologically-ordered sampling of rational subgroups, etc.) the calculated value is mostly useless.
 
Top Bottom