Additional information re GR & R - I was managing a mechanical testing laboratory & our requirements for measurement acuracy were more rigorous than in other areas of the mill. Device repeatibility was in proportion to cost of the device - the "cheapy" digital micrometers were not as repeatable as the more expensive ones. Fowler was good, Starret was good, offbrands were not as good. (Both Starret & Fowler met our requirements in the lab for required total accuracy.) Actually, I should probably not have used calipers as an example - we didn't do the comparison on calipers we did it on micrometers & then extended the conclusions across the board.
The devices were used in different parts of the mill, for the similar purpose of measuring sheet thickness, but environmental conditions under which the test were performed were very different - I got stuck doing additional GR & R's because I was the lab, and because other departments had run similar studies and had shown more measurement issues. The question arose as to whether the instruments used had something to do with the different results from the different departments. I received 2 additional brands of micrometers (1 Starret & 1 off-brand) from other sections of the mill and used the same samples and operators with the new instruments to run the additional GR & R studies. The Starret gave similar repeatibility to the Fowler I used initially. The off-brand was not as repeatible. Operator reproducibility was similar for all 3 brands with the lab operators.
We used the average & range method, with 1 instrument, 10 samples & 3 appraisers - this permits breaking down the results into repeatibility and reproducibility. Under lab conditions reproducibility was better than under process conditions in the mill - we didn't delve deeper, but it wasn't surprising - considering environment, training, etc. differences between the 2 locations & sets of employees making the measurements.
Hope this clarifies matters somewhat.