Minimum sample size - Guidance and statistical rationale

[MasterBB]

looking for guidance and statistical rationale sources using a sample size of 9, 12, or 15.
This is due to cost, timing, etc.

The units are a medical device but not life-threatening.

Can you advise of an appropriate sample size for Process validation?

I am not looking for a sample size based on Reliability & Confidence level and thus avoiding n = 29.

Any help appreciated.

 

[gogreen]

There are some commonly used studies that help determine the number of participants you should have. Have you tried searching around?

 

[Bev D]

First let me say that no validation sample sizes should be driven by cost. That is just incredibly short sighted and wrong. Period.
Given the severity of the potential failure modes, one can adjust the study designs to determine the appropriate sample sizes commensurate with the severity of the failure modes.

Without any additional information on the process we cannot logically tell you the appropriate sample size.

 

[Steve Prevette]

Agreeing with Bev. The sample sizes you are proposing would be considered too small by any regulator or statistician. One thing to consider - if this is a continuing product, and you sample a few each month, you will eventually build up sufficient sample size to have confidence in the item. This is how MIL-STD-105 and ISO successor sampling plans operate.

Also, if you are sampling by a measurement rather than go-no go, sometimes (depending on the standard deviation that goes into the student-t distribution) you can get by with fairly small sample sizes.