Missing Data in Discrete Time Surviva... PreviousNext
Mplus Discussion > Categorical Data Modeling >
 Sarah Victor posted on Tuesday, July 31, 2018 - 8:23 am
Hello all,

I am working on a discrete time survival analysis using mixture modeling as per the Muthen & Masyn 2001 paper. My current model appears to be kicking out participants with missing data on any covariate, which I do not want; I would like to bring those variables into the model by estimating their variance.

Here is the syntax:
USEVARIABLES SH14 SH15 SH16 SH17 cov1 cov2;
CLASSES = c(1);
f BY SH14@1 SH15@1 SH16@1 SH17@1;
f ON cov1 cov2;
!cov1 cov2;
f ON cov1 cov2;
!cov1 cov2;

When I try to add the lines commented out above, I get the following error:

One or more MODEL statements were ignored. Note that ON statements must appear in the OVERALL class before they can be modified in class-specific models. Some statements are only supported by ALGORITHM=INTEGRATION.

When I add ALGORITHM=INTEGRATION to the ANALYSIS code, it runs fine. Is this appropriate for use in mixture modeling with discrete time survival analysis?
 Sarah Victor posted on Tuesday, July 31, 2018 - 8:40 am
I should perhaps also specify that adding ALGORITHM=INTEGRATION allows the syntax to run when I assume proportionality (using my SH variables as indicators on a latent factor), but will not run when I assume non-proportionality; in that case, I get an error saying that I need to use Monte Carlo estimation. In that case, adding the line INTEGRATION=MONTECARLO allows it to run, but again, I am not sure if this is appropriate for this type of analysis.
 Bengt O. Muthen posted on Tuesday, July 31, 2018 - 6:02 pm
Int=MonteCarlo is appropriate for the analysis.
Back to top
Add Your Message Here
Username: Posting Information:
This is a private posting area. Only registered users and moderators may post messages here.
Options: Enable HTML code in message
Automatically activate URLs in message