RDSEM with single-indicator latent va... PreviousNext
Mplus Discussion > Dynamic Structural Equation Modeling >
Message/Author
 Andrew Johnson posted on Tuesday, July 07, 2020 - 6:43 pm
I'm attempting to run an RDSEM model with single-indicator latent variables.

As a first step, I ran a model with the latent variable loading fixed to 1 and the residual variance fixed to zero, assuming that the results would be identical to an RDSEM with observed variables. However, the estimates are all markedly different and the DIC is too large to be displayed (all asterisks).

I've copied the code below, does it look like I've missed something? Thanks!

VARIABLE:
NAMES = ID TIME x;
MISSING = ALL(-99);
CLUSTER = ID;
USEVAR = x TIME;
LAGGED = x(1);
WITHIN = TIME;

ANALYSIS:
TYPE = TWOLEVEL;
ESTIMATOR = BAYES;
FBITER = 5000;
CHAINS = 2;

MODEL:
%WITHIN%
F_x BY x@1 (&1); x@0;
F_x ON TIME;

F_x^ ON F_x^1;

%BETWEEN%
x;
 Bengt O. Muthen posted on Wednesday, July 08, 2020 - 11:53 am
In Bayes, you don't want to fix a variance exactly at zero. Say

x@0.01;
 Andrew Johnson posted on Wednesday, July 08, 2020 - 7:19 pm
Thanks Bengt, unfortunately that did not fix the issue. I'll send the input and data through to support now.
Back to top
Add Your Message Here
Post:
Username: Posting Information:
This is a public posting area. Enter your username and password if you have an account. Otherwise, enter your full name as your username and leave the password blank. Your e-mail address is optional.
Password:
E-mail:
Options: Enable HTML code in message
Automatically activate URLs in message
Action: