It's a good ones Dave, I agree that the traditional R&R uses Control Charting as it's base, so if the distribution is non gausian the control limits and statistics could be mislead, but....
In an R&R, the most important issue is the difference between operators, and IMO not seriously affected by a non gausian behabiur of the data (because the data points of each operator is in the same range).
I really don't have information about non normality and R&R but as an experim ent you can normalize the data using the transformation method you normally used to normalize the data an see what happend.
Give us a feedback of the results, I tink that we all are interested on the outcome.
