How many samples

Home > Statistics Every Writer Should Know > The Stats Board > Discusssion

How many samples
Message posted by stuart (via 203.96.111.202) on May 14, 2001 at 4:16 AM (ET)

Hi, we manufacture several 4 products. I have been asked how many samples of final product do i need to check on a monthly basis to be 95% confident that there are no more than 1% defective products in any month. Typical monthly output is 6000 of 2 products and and ~500 of the other 2. The check list would look at several chacteristics...thanx,

READERS RESPOND:
(In chronological order. Most recent at the bottom.)

Re: How many samples
Message posted by mbd (via 205.188.197.187) on May 23, 2001 at 8:30 PM (ET)

If you want to be sure there's no more than 1% defective items in the product, then you need to improve your process. No sampling method can reduce the number of faulty units in a lot. The quality is already there.

A better question is "How large a sample do I need to estimate the proportion defective within x?" Refer to a statistics text...look for "Required Sample Size."

Re: How many samples
Message posted by Jack Tomsky (via 208.249.113.130) on May 25, 2001 at 5:52 PM (ET)

For each product and each month, suppose that your acceptance number is zero. This means that you will accept the product's quality for that month provided you observe zero defectives out of N and reject the product if you observe one or more defectives.

Your question is what should the sample size N be. To have 95% confidence of rejecting the product when p is 1% or more,

N = ln(.05)/ln(.99) = 298.

The mathematics can be worked out for cases when you choose a different acceptance number, greater than zero. That would lead to larger sample sizes.

Your $5 contribution helps cover part the $500 annual cost of keeping this site online.