Hi Everyone,
After a quick investigation both internally and with our suppliers it has come to my attention that very few people actually understand the results of a gR&R study and what to do with the results so I have made a training presentation which is aimed at quality engineers who are responsible for supplier quality improvements. Its main aim is for them to understand why, what and how to perform a GR&R and how to interprete the results (not just looking at the AIAG % and not understanding it) I would like them to make informed decisions on the suppliers measurement systems and what needs to be done if anything to improve any measurement systems.
So could any of you statistical experts and GR&R experts please
1. - check the presentation for any errors in assumptions, explanations etc.
2. - Check the 2 excel files attached for any errors.
3. - Make suggestions for improvement.
Anyone is free to use this presentation if they want.
Thanks in advance
Simon
Thanks for sharing this.

I think it'll be very useful for a lot of people here.
Below are a few observations that might help to improve it. The text in bold is quoted from your slides.
Slide 4:
Since you cannot address something that cannot be measured precisely or the when the measurement precision is unknown, you must start with an assessment of the measurement system.
This is a bit vague; if something
cannot be measured precisely, it cannot be measured precisely, and GR&R won't help.
Slide8:
A measurement error is not a "mistake". Variability is an inherent part of things being measured and of the measurement process.
Measurement error
can be a mistake on the part of an operator.
Slide 12:
Repeatability – variation of the gage when used by one operator in a brief interval.
The "brief interval" bit is irrelevant. It should be assumed that repeatability problems in the study will be projected out to the actual process as time passes.
Slide 15:
The study must be conducted in the same area, environment, fixtures and conditions that will be used in production.
"Must" is a bit strong; you're describing the
ideal conditions, and those conditions aren't always feasible.
Slide 23:
There are 2 main methods to calculate Gage R&R are Average & Range and the other is ANOVA. The difference is that ANOVA analyse any interactions between the repeatability and reproducibility.
Grammar issues. It would be better stated as, "The two primary methods of GR&R analysis are
Average and Range and ANOVA. The latter is used to analyse interactions between repeatability and reproducibility."
Slide 50 (Common Mistakes):
(8) Using the result of 1 Gage R&R study on 1 characteristic to read across to other parts or characteristics that use the same or same type of gage.
It's considered acceptable practice to apply the results of a study to other parts with characteristics that are similar in size and geometric configuration.
(9)
Conducting the study in Lab conditions and not at the station where the measurements are actually taken in production.
See my comment regarding slide 15 above.
(10)
Not understanding and doing nothing with the results.
These are two separate categories. Not understanding the results is certainly bad, and what one does with the results is dependent on the information provided by the results. Sometimes the results indicate that the suitability of the measurement system has been confirmed, in which case doing nothing might be the proper strategy.
(11)
Not repeating the Gage R&R study on a minimum 12 month basis.
Repeating a gage study shouldn't be necessary unless there have been changes that would make it sensible, such as new operators or a significant change in the operating environment. If you've confirmed that the measurement system is appropriate to the task, repeating it won't help anything.
Slide 53:
Studies of measurement variation are a waste of time unless you use that information to reduce process variation, increase process control and assure product confirmation.
As suggested above, sometimes simple confirmation of the efficacy of the measurement system is enough. It shouldn't be considered a waste of time if the results aren't used to reduce variation or increase process control, both of which might already be acceptable.