Latent Class with Covariates, R3STEP PreviousNext
Mplus Discussion > Categorical Data Modeling >
Message/Author
 Georg Datler posted on Tuesday, June 28, 2016 - 8:09 am
Hello,

I used the R3STEP.
Now I do not get the PLOT "Estimated probabilities for a categorical latent variable as a function of its covariates"
anymore.

Is there a way to get it? (If not, could you point me to the reason?)

Many thanks
Georg Datler
 Bengt O. Muthen posted on Wednesday, June 29, 2016 - 6:51 am
We do give that plot also with R3STEP so there must be something else going on with your run - to diagnose it we have to see your output. Send to Support along with your license number.
 Danyel Moosmann posted on Friday, July 13, 2018 - 1:04 pm
Hello,
I ran a longitudinal LCA and have selected a 4 class model and now want to look at covariates. Can you tell me how Mplus handles missing data on the indicators when using R3step? My indicators do have missing data, but when I ran the unconditional model, I used FIML and all of my cases were used. However, when I ran the conditional model, I lost several cases. Is there a special way to invoke FIML when using R3step?

Thanks so much.
Danyel
 Tihomir Asparouhov posted on Friday, July 13, 2018 - 3:15 pm
The reason for the deleted cases is missing data in the covariates (not the indicators). You can use the manual R3step, see section 3
http://statmodel.com/download/webnotes/webnote15.pdf
where you can include a model for the covariates. Instead of
C on X; you would use C on X; X;
 Danyel Moosmann posted on Friday, July 20, 2018 - 4:21 pm
Thanks so much, Tihomir.

Danyel
 shanting chen posted on Friday, July 19, 2019 - 7:18 am
My question is regarding FIML and dealing with missing data when adding covariates to an LCA model. Has there been any new developments regarding handling missing data when using the automatic R3STEP (I believe the default is listwise deletion), or the recommended method still the manual R3STEP to avoid losing the cases that have missing in one or more of the covariates?
 Bengt O. Muthen posted on Friday, July 19, 2019 - 1:44 pm
Still the same.
 aprile benner posted on Tuesday, August 27, 2019 - 7:25 am
To avoid listwise deletion, we are trying to conduct a manual R3STEP analysis for a model in which we are regressing a latent variable (c) on covariates. Webnote 15 details how to regress a distal outcome on a latent variable with very useful syntax. I cannot, however, locate example syntax for our analyses, and we are having difficulty adapting the syntax from Webnote 15. Is there available sample syntax for the manual R3STEP that includes a latent variable regressed on covariates?
 Bengt O. Muthen posted on Wednesday, August 28, 2019 - 2:48 pm
When you say latent variable in your last sentence, do you mean c? I assume you have seen this document that goes with Web Note 15:

http://www.statmodel.com/download/AppendicesOct28.pdf
 Chris Greenwood posted on Monday, September 21, 2020 - 9:16 pm
Hi, I have followed the manual 3-step approach to first estimate an LCA and then estimate the effect of a covariate on the classes. My covariate has missing. I can replicate the automaic 3-step version when i ignore the missing. However, when I bring the covariate into the model I get an error:

*** ERROR
The following MODEL statements are ignored:
* Statements in the OVERALL class:
C#1 ON AGE
C#2 ON AGE

This is some of the code:
USEVARIABLES ARE
Age n ;
NOMINAL ARE
n ;
CLASSES = c(3);

ANALYSIS:
TYPE=MIXTURE; STARTS=0;

MODEL:
%Overall%
c ON Age; Age;

%C#1%
[N#1@3.431];
[N#2@1.246];

%C#2%
[N#1@1.676];
[N#2@4.439];

%C#3%
[N#1@-3.644];
[N#2@-3.749];

I have tried a few alternatives using the AUXILIARY variable command, but they dont allow me to bring the covariate into the model.

Any help would be greatly appreciated.
 Chris Greenwood posted on Monday, September 21, 2020 - 11:21 pm
In regards my above post, this was solved by adding

ALGORITHM=INTEGRATION;
INTEGRATION=MONTECARLO;

Is this ok?
 Bengt O. Muthen posted on Tuesday, September 22, 2020 - 10:07 am
See the new Section 7.1 of web note 21 which deals with missing on Xs.
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: