How JMP handles extreme outliers in the capabilty analysis?

Question

Does the capabilty analysis takes into account extreme outliers?

It’s a problem for Johnson and Normal distributions. As I see the problem JMP uses the extreme points for Goodness of Fit test, but not for Capability. See example attached

Data ark til analyser_ 2.jmp

malcolm_moore1 · Answer

What is the issue here? The capability calculations based on Johnson Su seem to be giving the correct guidance to me. The process has 5.5 sigma capability, indicating we should expect to see very occassionally out of spec results (over a muuuuch bigger time period). If needed there is the opportunity to improve further by centering the distribution and identifying the casue of the outliers. The process is already very good and the decision to improve further may need to be based on balancing the effort of further improvement vs benefit.

Best,

MikeD_Anderson · Answer

You would either want to screen them (if they have assignable cause that has been addressed) or use Non-normal Capability calculations in the process capability platform.  It’s been my experience that if they have a lot of extreme outliers their process is probably not normal.

Best,

M