Auto-suggest helps you quickly narrow down your search results by suggesting possible matches as you type.

- JMP User Community
- :
- Discussions
- :
- AICc log likelihood - where is it reported?

- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page

Highlighted

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

Created:
Jun 10, 2020 8:12 PM
| Last Modified: Jun 11, 2020 10:03 AM
(681 views)

Anyone know if/where the the -2LogLikelihood AIC values are reported when conducting model selection in JMP 14?

The "Likelihood, AICc, and BIC" page describes what I want to do:

- You can use the value of negative log-likelihood to choose between models and to conduct custom hypothesis tests that compare models fit using different platforms in JMP. This is done through the use of likelihood ratio tests. One reason that -2LogLikelihood is reported in many JMP platforms is that the distribution of the difference between the full and reduced model -2LogLikelihood values is asymptotically Chi-square. The degrees of freedom associated with this likelihood ratio test are equal to the difference between the numbers of parameters in the two models (Wilks 1938)."

I've attached a screenshot of the parameters and statistics I see in the model selection window.

Thanks!

2 ACCEPTED SOLUTIONS

Accepted Solutions

Highlighted

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

UPDATE: I've figured it out; I needed to click "Make Model" and change the personality type.

Highlighted

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

You are right - JMP does not report the -2L for the fitted model. It only provides the AICc and the BIC, which are used for model selection.

You can compute -2L = N * Log( 2 * Pi * SSE / N ) where N is the sample size, Pi is irrational constant, and SSE is the model-dependent estimate of the sum of squares error.

Learn it once, use it forever!

4 REPLIES 4

Highlighted
##

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

Re: AICc log likelihood - where is it reported?

Click on the "Run Model" button, then in the Fit Group Window, click on the red triangle and select

Regression Reports==>AICc

Jim

Highlighted
##

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

Re: AICc log likelihood - where is it reported?

Thank you for your response.

On my end, the regression report > AICc only provides the AICc value, which

was also provided in the model selection window. I'm looking for the log

likelihood itself, one component that goes into the determination of the

AICc value.

For example, this discussion post

mentions

"JMP invariably reports positive LLs and negative AICs, for example "-2LL =

-1406.88" or "AICc = -1451.534". I'm wondering where they found the "-2LL"

(-2log likelihood) number.

On my end, the regression report > AICc only provides the AICc value, which

was also provided in the model selection window. I'm looking for the log

likelihood itself, one component that goes into the determination of the

AICc value.

For example, this discussion post

mentions

"JMP invariably reports positive LLs and negative AICs, for example "-2LL =

-1406.88" or "AICc = -1451.534". I'm wondering where they found the "-2LL"

(-2log likelihood) number.

Highlighted

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

UPDATE: I've figured it out; I needed to click "Make Model" and change the personality type.

Highlighted

- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Get Direct Link
- Email to a Friend
- Report Inappropriate Content

You are right - JMP does not report the -2L for the fitted model. It only provides the AICc and the BIC, which are used for model selection.

You can compute -2L = N * Log( 2 * Pi * SSE / N ) where N is the sample size, Pi is irrational constant, and SSE is the model-dependent estimate of the sum of squares error.

Learn it once, use it forever!