Yeah- my engineers try that with me as well sometimes.
Then I run the GRR and in analyzing the results find out that with and without the parts in the tester it will still run above and below spec...so how can you use it to determine if the part is good or not if the variation exceeds the limits?
My point being that if the measurement equipment is less accurate/precise than the size of the thing being measured - the error bar on the thing being measured doesn’t account for everything or assumes that all the variation is in the thing being measured, so all the relationships are screwed up.
OK, noted. I’m commenting from a personal perspective (obviously) going back to my time in experimental nuclear physics - solid state physics using nuclear methods, to be completely honest, but my particular work group was part of the nuclear physics department. Anyway, the papers we submitted for publication always took - or at least tried to take - into consideration the instrumental errors as well as the measurement errors. Reflected by error bars in the data offered for publication.
Perhaps I’m too far removed from the field by now (having switched to a career in patent law) to understand what other errors you might refer to.