About loglikelihood PreviousNext
Mplus Discussion > Structural Equation Modeling >
Message/Author
 gibbon lab posted on Friday, December 10, 2010 - 12:13 pm
Every time I run a SEM, there is a loglikelihood for H0 and H1. For example,

Loglikelihood
H0 Value -13701.249
H1 Value -13697.746

What exactly are the null hypothesis and alternative?
 Linda K. Muthen posted on Friday, December 10, 2010 - 12:45 pm
The H0 model is the model specified in the MODEL command. The H1 model is the unrestricted means, variances, and covariances of the observed variables.
 gibbon lab posted on Friday, December 10, 2010 - 1:07 pm
Usually we do not want to reject the specified model. Does that means we are looking for non-significant p values for the Chi-Square Test of Model Fit?
 Linda K. Muthen posted on Friday, December 10, 2010 - 1:50 pm
Yes.
 almoskowitz posted on Monday, March 07, 2016 - 10:54 pm
Hello!

I've requested for MPlus to save the individual loglikelihoods using the SAVEDATA; SAVE=LOGLIKELIHOOD; function and it seems to be outputting the log likelihoods commensurate with H1 rather than the fitted model H0.

Is there any way to get the log likelihoods from the fitted model instead?

Thank you!
 Linda K. Muthen posted on Tuesday, March 08, 2016 - 10:06 am
Can you send whatever makes you think they are from the H1 model to support@statmodel.com.
 Michael Hallquist posted on Tuesday, January 30, 2018 - 8:03 pm
Hi Mplus gurus,

Apologies if this is a naive question, but why does the unrestricted (H1) model LL change when there are exogenous covariates in the model compared to a model without covariates? My understanding is that the H1 represents the variance-covariance matrix of the endogenous variables.

Does this change because the endogenous variables are conditioned on the covariates and thus the unrestricted model is the covariance of the residual variance?

Thanks much,
Michael
 Bengt O. Muthen posted on Wednesday, January 31, 2018 - 4:39 pm
The answer is yes. You can also state H1 as

Y = a+bX +e
Var(e)=Sigma

where a, b and Sigma are unrestricted.
 Bharath Shashanka Katkam posted on Wednesday, August 21, 2019 - 6:58 am
Hello Mplus Team,

I ran a saturated Model in the Mplus, with the following code:

Model:

%Within%
Y1 with Y2-Y4;
Y2 with Y3-Y4;
Y3 with Y4;

%Between%
Y1 with Y2-Y4;
Y2 with Y3-Y4;
Y3 with Y4;

when I ran this model, it has shown the Loglikelihood values

H0 Value -4906.448
H1 Value -4905.948

Q) My question is, when we are running a Saturated (Unrestricted) Model, why would it show two different values of H0 & H1?
I feel that, the Unrestricted Model (H1 value) & test model (H0 value), which is also Unrestricted (i.e., Saturated), should have the same value...
Why are the both values not same?
 Bengt O. Muthen posted on Friday, August 23, 2019 - 6:17 am
They should agree. Try sharpening the convergence criteria - see the Summary of Analysis section in the output to see what the default criteria are.

If this doesn't help, send your output and data to Support along with your license number.
 Bharath Shashanka Katkam posted on Saturday, August 24, 2019 - 7:13 am
Okay Sir. Thank you
Back to top
Add Your Message Here
Post:
Username: Posting Information:
This is a private posting area. Only registered users and moderators may post messages here.
Password:
Options: Enable HTML code in message
Automatically activate URLs in message
Action: