This website uses Cookies. Click Accept to agree to our website's cookie use as described in our Privacy Policy. Click Preferences to customize your cookie settings.
Accept
Reject
Preferences
cancel
Turn on suggestions
Auto-suggest helps you quickly narrow down your search results by suggesting possible matches as you type.
Showing results for 
Show  only  | Search instead for 
Did you mean: 
JMP® 15 Fitted Distribution results might differ from results in previous versions of JMP
Created: Nov 30, 2022 02:33 PM  |  Last Modified: Dec 20, 2022 03:58 PM

The JMP 15 Distribution platform was modernized to have distributional fitters that match what is available in the Generalized Regression platform. Differences in computational methods can result in differences in reported confidence intervals, the best fitting distribution, and so on, between JMP 15 and earlier versions of JMP.

In addition to some newly available distribution fits, additional features for existing distribution fits, an improved Compare Distributions report, usage of the Process Capability platform for capability analyses, and much more, some notable differences are the following:

  • In the old fitters, sometimes Likelihood confidence intervals are used, and other times Wald confidence intervals are used. The new fitters in JMP 15 are standardized on using Profile Likelihood confidence limits for parameters, with the exception of the Johnson and Normal Mixture Distributions where Wald intervals are used because the profile limits aren’t feasible.
  • For the Fitted Normal distribution, specifically, JMP 15 computes confidence intervals based on the Z statistic, whereas previous versions use a t-statistic and matched the Summary Statistic report output.
  • The Goodness of Fit test has been standardized to the Anderson-Darling test for continuous fits and Pearson Chi-Squared for discrete fits. Previously, the Goodness of Fit test differed based on the type of distribution and/or the sample size. (See the table here.) Note that the Anderson-Darling test is simulation-based, so the exact same result might not be produced with each run.

If you want to have matching results between JMP versions, enable the Legacy Fitters in the JMP 15 Distribution platform by selecting the red triangle menu next to your analysis column and then selecting Continuous Fit ► Enable Legacy Fitters.

For more comprehensive information about the modernization of the JMP 15 Distribution platform, see the following blog posts:

Modernizing the Distribution platform – New fitters in JMP 15
Modernizing the Distribution platform – Comparing distributional fits in JMP 15
Modernizing the Distribution platform – Options for fitted distributions
Modernizing the Distribution platform – Capability analysis

 

[Previously JMP Note 66137]

0 Kudos
Details
Operating System
macOS Windows
Reported Version
15.0
Fixed Version
15.0
Products JMP JMP Pro