Dude,
I hope this analogy explains it without telling you something you don’t already know.
It’s like a right triangle obeying Pythagoras’s theorem: the sum of the squares of the legs equals the square of the hypotenuse. Standard deviations are the lengths of the legs and variances are the squares of the lengths of the legs.
So, assume the legs represent measurement and part variation. The hypotenuse is total variation. For ease of analogy, let’s say the measurement leg (GRR) is 3, the part-to-part leg is 4, and the hypotenuse is 5 (a 3-4-5 right triangle).
Using StudyVar: GRR is 3/5 or 60%. Part-to-part is 4/5 or 80%.
Using Contribution: GRR is 9/25 or 36%. Part-to-part is 16/25 or 64%
Note that Contribution is the square of the SudyVar (0.6^2 = 0.36). I don’t know why yours didn’t come out that way; I’d need to see the whole data set. Maybe you’re mixing StudyVar and %Tolerance.
So that’s all it is. With contribution, the percentages will add to 100. With StudyVar, they won’t. But StudyVar is in the natural units of measure, and Contribution uses the square of them. It’s your choice; whichever you prefer. Auto industry uses StudyVar, because standard deviations are in the natural units.
.
.