Creation of Field Test Plan and sample size

[John Adamo]

I am developing a plan to field test(actual customer sites) a new product to assure that the customer expectations will be met. We know that there are a number of variables which will effect the performance of the product. For conversation sakes, let us assume that there are 4 factors.

The question is how many samples would we need to put into the field to demonstrate a 95% reliability (with a 90% confidence).

Do you have other examples on how to approach field test?

[Dr. Dave Olwell]

John,

The number of samples required depends on the degree by which you exceed the requirement. For example, if you have a true reliability of 95.1%, it will take many more samples to establish 95% reliability with 90% confidence than if your actual reliability is, say, 99%. We cover this in our seminars, albeit briefly.

Since the answer depends on the underlying reliability, we can help with a sequential testing approach. A small screening test allows you to fit a preliminary model that in turn can be used to estimate the sample size needed for the full experiment and desired precision.

If you'd like help with that, contact ReliaSoft Professional Services.

An alternate approach is to make assumptions about your underlying reliability, and then design the test. That, of course, is only as good as the assumptions.

Best wishes,

Dr. Dave