Forum for Sampling Practices

Are my statements below correct?

AQL
-Defective % below the AQL are routinely accepted by the plan.
- If I fail the sampling plan, I am 95% confident that the % defective is above the AQL. (assuming alpha is 0.05)


RQL (or LTPD)
-Defective % above the RQL are routinely rejected by the plan.
- If I pass the sampling plan, I am 90% confident that the % defective is below the RQL. (assuming beta = 0.10)

ed adams wrote:Are my statements below correct?

AQL
-Defective % below the AQL are routinely accepted by the plan.
- If I fail the sampling plan, I am 95% confident that the % defective is above the AQL. (assuming alpha is 0.05)


RQL (or LTPD)
-Defective % above the RQL are routinely rejected by the plan.
- If I pass the sampling plan, I am 90% confident that the % defective is below the RQL. (assuming beta = 0.10)

Yes, your statements are correct.

The description of what sampling plan decisions mean can take either of two forms -- hypothesis testing and confidence statements.

The hypothesis testing description, with its logical "if-then" form, parallels the way the sample size equations are derived.

The relationship between confidence intervals and hypothesis tests is that any hypothesis that you state within the interval cannot be rejected by a hypothesis test that uses the same data. On the other hand, any hypothesis outside the interval would be rejected by that data.

For the case at hand we have a one-sided upper confidence interval and a one-sided upper hypothesis test.

The confidence statement description does not parse as well but it communicates more easily. The only problem that I have seen (too often) with the confidence method of explaining is that a large number of people think in either/or terms and interpret (wrongly) that "below RQL" means the same as "is AQL". It's a good topic for the dinner table at ASQ meetings.

Stan Hilliard