Sample Size to Test Products used for 100 hours in a Test Environment

[philingle]

Hi

Sorry if this question has been asked before. I have looked around and cant find a simple answer to it here.

I have some products for which there is a european standard, the products need to withstand static loading and dynamic loading.

I have a batch of 25 products that have been in service for 100 hours in an environment where they have been exposed to a mild acid. I want to test those products to see whether after 100 hours exposure they are still up to the standard.

How many of them do I need to test to prove that they remain up to standard? Is it 100% of them, or can I test a smaller number?

Thanks

Phil

[Bev D]

well it depends.

it depends on the characteristic you are measuring (continous/measurement data or categorical/pass-fail data) and how much variation these products experience for the characteristics of interest. Do you know the standard deviation from historical data?

it also depends on if ALL of the values must be within standard or only the average...the common equations are for estiamting the average which are often not useful for these types of tests because we are interested in the individual values.

can you tell us a little more about what you are doing?

[Steve Prevette]

Most likely you better test all. If I interpret your question correctly - you have put 25 units through a 100 hour test, and now you want to know how many need to be checked to see if they are satisfactory.

The thing is, even if you tested all 25, and they all came up satisfactory, at best all you could say is that you are 92% confident that at least 90% of the product are satisfactory after a 100 hour test.

Most organizations look for 95% confidence that at least 95% are good - and that takes 59 tests without failure to prove.