Hi HarryYin
Given the tag, I guess that you are talking about GRR studies. However, there are several statistics which are calculated during these studies.
Equipment variation
Appraiser variation
Part variation
GRR
Total variation
Which of these are you referring to?
As to which you should use:
"
Either or both approaches can be taken depending on the intended use of the measurement system and the desires of the customer" (MSA manual, 4th edition, page 123).
Neither approach is the "correct" approach. They simply give you different comparisons. If you compare to process, you get a measure of your ability to detect what your process actually produces.
If you compare to tolerance, you are only looking at your ability to detect if a product is within specification.
You may think that the two statements are equivalent, but in the real world, process capability and tolerance can be very different.
Let's say you have a completely lousy manufacturing system where most of your items are either way too big or way too small. You can still have acceptable product going out of the door.
How?
Easy! Just have 100% inspection with a measurement system that can just about tell the difference between 'in tolerance' and 'out of tolerance'. Then you throw away all the useless parts. You may have a scrap rate of 99%, but your customer only gets good parts.
On the other hand, you have a truly awful manufacturing system.
Now carry out a GRR study on this setup. If you compare your barely capable measurement system to the product tolerance (what the customer requires), you will find it is barely capable.
If you now compare your barely capable measurement system to the truly awful manufacturing system, you will find that your measurement system is amazingly good.
Actually, your measurement system is barely capable, you are just comparing it to a production variability that is so great you may as well sort your initial output using a couple of industrial sized seives. ("If it falls through these holes, it's too small, if it doesn't fall through those holes, it's too big"

)
NC