What inspired this wish list request?
Found some inaccuracies with the case of non-normal part variation and test/re-test variation on probabilities of misclassification.
What is the improvement you would like to see?
Add support for Gauge study analysis (including the misclassification probability calculation piece) with a nonparametric or semi-parametric modeling approach. This may or may not include support for bootstrapping on the Misclassification Probabilities report outline.
Why is this idea important?
This may increase various accuracies of estimates when part-to-part and measurement variation are severely not-normal (but still considered stable) when studied in a Gauge R&R.