How to determine sample size for known defect size?

Aravindan

We are currently testing some chips and have a high open short failure. We are taking a sample size of 60% of the original lot and performing a sample test. Our unaccepatable defect rate for whole lot is 0.3%. I want to know with a lot size of 8000 or 10000 and a sample size of 60% of this lot size. What will be my defect rate should be to determine my whole lot has a defect rate of 0.3%. 

 

Example

Lot Size : 8000 or 10000

Sample size: 60% of Lot Size

How do determine whats my sample defect rate which will indicate that my lot has 0.3% defect rate.

 

A JMP or Excel formula instruction will be useful

Mark_Bailey

A sample size determination is made in the context of a statistical hypothesis test. This context means that you have to specify the hypothesized proportion and the smallest change in the proportion that is important to you. You also have to specify the desired power in the test. That is, the probability that you will find the difference to be significant when in fact the true difference is at least as large as you specified.

1. Select DOE > Design Diagnostics > Sample Size and Power.
2. Click One Sample Proportion.
3. Enter the significance level desired (default is alpha = 0.05).
4. Enter the hypothesized proportion, 0.003 (for 0.3%) in this case.
5. I would use the default estimation method.
6. Select one-sided if you are interested in an upper bound or two-sided if you want an interval.
7. Leave the Null Hypothesis empty.
8. Enter 4800 for the Sample Size (60% of 6000).
9. Enter 0.9 for the Power.
10. Click Continue.

Here is the result:

This means that you have 90% of finding a change to 0.59% failures with this sample size. Changing the sample size to 6000, the null proportion decreases to 0.57%:

See JMP Help or this white paper for more details.