Gage R&R on Transformed Data?
December 4, 2017 at 8:53 am #55889
In my Gauge R&R analysis (4 parts, 4 operators, 3 repeats), the plot of residuals versus fitted values exhibits a funnel shaped pattern; the variance increases as the measured response increases. A square root transformation on the response variable does a nice job of correcting this problem.
But…is it ok to perform GRR on transformed data? My concern is that the interpretation of the results are now based on the square root of the measurements. Can I still make a conclusion on the repeatability and reproducibility of the measurements based on their square root? The actual variance component values are smaller, but the %contribution remains relatively unchanged (compared to non-transformed data).
Thank you!December 4, 2017 at 10:29 am #202030
Did your gage pass? Transformation is a neat tool but if the outputs/inputs aren’t precisely measured by the gage, transformation won’t help.
You must be logged in to reply to this topic.