MSA is probably one of the least understood and one of the most abused of investigation methods on reliability of measuring systems: I've seen it carried out on destructive measurements, by laboratory technicians instead of line operators only because Registrars' auditors request it. Even when the system is evidently stable, third party auditors request MSA be carried out once a year, at a company's great expense of time and money. And we are talking of auditors as graduated engineers, not of common laymen. It is therefore no wonder that companies' metrologists seldom care for it: once the output is within specs, it's cooked and ready for serving.

I have a book from Dr. Wheeler Evaluating the Measurement Process III and I fully agree with his opinion about percentages of acceptability of measurement system. In short these percentages are deteriorating the measurement system. If the measurement system is stable, predictable and consistent validated by process behavior charts, I will not say that MSA with R&R = 35% is unacceptable. Depends on use. One of the purpose of measurement system is to be able to track process changes and for that is MSA with R&R = 50% still enough. I work in a machinery company with lots of single-purpose process measures and with this approach we should scrap half of them and then stop the lines. Unfortunately our customers request this approach because they are widely using AIAG studies as a Holy Bible which is unfortunately wrong.

I propose to read Dr. Wheeler´s article "Problems with Gauge R&R Studies" here on Quality Digest for more information and for different view. http://www.qualitydigest.com/inside/twitter-ed/problems-gauge-rr-studies.html