[Bug]: Process Capability won't generate on 3-parameter Weibull data

[Lisa-Capybara]

JASP Version

0.96.0

Commit ID

No response

JASP Module

Quality Control

What analysis are you seeing the problem on?

Process Capability Study

What OS are you seeing the problem on?

Windows 10

Bug Description

Process Capability generates on normally distributed data as well as data that doesn't fit the 3-parameter Weibull. "This analysis terminated unexpectedly" whenever I apply process capability, 3-parameter Weibull, to data that in fact fits the 3-parameter Weibull. Additionally, the Anderson Darling always has a p of 0 with non-normally distributed data against the 3-parameter Weibull, even when I know it fits the 3-parameter Weibull based on Minitab output on same data.

3 Parameter Weibull errors.zip

Expected Behaviour

Based on Minitab output on same data, Anderson Darling against 3-parameter Weibull should be, AD: 0.313, p > .5 and there should be output generated with expected overall performance of 2.76% defect with Upper Spec Limit of 4.

Steps to Reproduce

With data based on the 3-parameter Weibull distribution, when I request Non-normal distribution, 3-parameter Weibull, I receive an error that "This analysis terminated unexpectedly. Error in stats::optim(x = c(1.5, 1.1, 2.1, 1.2, 1.9, 4.2, 2.2, 2.2, 1.4, : non-finite finite difference value [3]... Stack trace... to receive assistance with problem, please report the message above at: https: //jasp-stats.org/bug-reports."

Log (if any)

No response

More Debug Information

No response

Final Checklist

• I have included a screenshot showcasing the issue, if possible.
• I have included a JASP file (zipped) or data file that causes the crash/bug, if applicable.
• I have accurately described the bug, and steps to reproduce it.

[EJWagenmakers]

Important to check this out. Maybe we only use the AD test to test for normality? Plus important to check the fitting routine

[JTPetter]

@Lisa-Capybara Thank you for the report and data example. I looked into this and prepared a fix. The fitting function already returns a more informative error in the latest update of the module. In addition, I was able to make it more robust, such that it actually manages with your data now. Regarding the AD, note that it is indeed not implemented for the 3-parameter Weibull. Still, the p-value should not be 0, that was a formatting issue, which I also fixed.