Is testing 99.7% reliable? Reliability Question

[7u7u1]

Guys,

I have these questions asked by a customer. I do not have knowledge in reliability. Can anyone help please? Thanks!

1. is testing 99.7% reliable?
2. if 5 bags are tested and result is less than 20cfu how reliable is the batch?

[kgott]

If an airline advertised that they got you to your destination 99% of the time would you still fly?

If the biro you write with meet the manufacturers claims 50% of the time how important would that be to you?

Your question can only be answered by you and your customer. Are they bags of sand or they bags of brain surgeons?

[Stijloor]

In Reply to Parent Post by kgott

<snip>If the biro you write with meet the manufacturers claims 50% of the time how important would that be to you?

For our international Members:

In British English the word "biro" is often used as a generic term for any ballpoint pen.

Stijloor.

[Steve Prevette]

In Reply to Parent Post by 7u7u1

Guys,

I have these questions asked by a customer. I do not have knowledge in reliability. Can anyone help please? Thanks!

1. is testing 99.7% reliable?

Why would you believe it to be so? 99.7% happens to be the three standard deviation spread on the normal distribution. There is no reason to believe that an arbitrary testing scheme would match that. You will find some are worse, you will find some are better. The only way to determine is to put KNOWN GOOD items through the testing process and see how many are rejected, and KNOWN BAD items through the testing process and see how many are accepted.

2. if 5 bags are tested and result is less than 20cfu how reliable is the batch?

What is a cfu and how does it relate to the specifications for the batch? If it is some form of measurement of error, I'd need to know both the average and standard deviation of the data (and preferably, all 5 data values) before being able to characterize the error, and then I need to know the specifications to be met to even begin to work on reliability.