JASP - Standard deviation used in Process Capability Study part of Quality Control Module

Superfinisher

Hello,

My organization is looking to move away from Minitab and we have run into an issue. We primarily use the Normal Capability Six Pack function of Minitab and saw that there is a Process Capability Study in JAST and were excited to possibly use it. The problem I am having is that minitab and JAST is that while they look to match in many ways they give different values for the "within" process capability. The "total" results are fine, but the Cpk is quite different for the "within" calculation. After digging it appears that this could be driven by the standard deviation "within" being different between JAST and Minitab.

There is lots of documentation out there on Minitab and i know that it is using a pooled standard deviation and I am able to recreate their results by calculating it by hand. However, I can't find any information on how JAST is doing the calculation so that I can understand why their results differ. Has anyone else run into this? I am pretty new to JAST so is there a way for me to dig in and get a better idea of how it is doing its calculations in the background?

Any help would be greatly appreciated.

EJ

Dear Superfinisher,

I will attend some of our team members to this issue. Please note that we have set up a service company to help industries switch to JASP from Minitab: https://www.jasp-services.com/

Cheers,

E.J.

ReneKlerx

First of all, thank you for using JASP.

JASP is using for the within standard deviation the formulae based on R-bar/d2 (d2 is the bias correction factor) in the case if the X-bar & Range control chart is used for the confirmation of the stability of the process. This is an old fashioned method of the previous century when we didn't have software to calculate the standard deviation. Not recommended to use.

Therefore the JASP default to confirm stability is using the X-bar & s control chart. With that control chart automatically the within standard deviation is calculated and based on s-bar/c4 (c4 is the bias correction factor).

When the within standard deviation is based on the pooled standard deviation, it depends very much whether the bias correction is used or not. This is an option in Minitab, which was in the past not always selected.

The pooled standard deviation is preferred when you don't have equal sample sizes, which is quite uncommon but it might be possible. In all other cases the difference between s-bar/c4 and s-pooled are not major.

Suggestion to JASP: it would be very beneficial to add an option to JASP to give the possibility to use the pooled standard deviation, as well.

Regards,

René