I am trying to run a random intercept cross-lagged panel model (cf. Hamaker, Kuiper, & Grasman, 2015, http://dare.uva.nl/document/2/168970). However, this model was designed with continuous observed variables, and my variables are non-normally distributed ordinal categorical variables. I tried to run this model without specifying the variables as categorical and that went fine. However, when I add the "categorical = ..." command, I receive an error message: "The following MODEL statements are ignored: * Statements in the GENERAL group: [ W1CBV ] [ W2CBV ] [ W3CBV ] [ W1CBP ] [ W2CBP ] [ W3CBP ]". This refers to the following line in the MODEL-command: [w1CBV-w3CBP@0];
With continuous variables, the goal is to fix the observed means to zero so that the latent means can be estimated. However, this does not seem to work with categorical variables. How can I run this model with categorical variables?
Thanks for the suggestion. I tried this, but received the following error message: THE STANDARD ERRORS OF THE MODEL PARAMETER ESTIMATES COULD NOT BE COMPUTED. THE MODEL MAY NOT BE IDENTIFIED. CHECK YOUR MODEL. PROBLEM INVOLVING THE FOLLOWING PARAMETER: Parameter 11, [ MCBV ]
From the tech1-output, I see that parameter 11 is the alpha-value for MCBV. In tech4, the means of MCBV is estimated to be 0.194, so I do not see what the problem is here.
Hi I would like to know if it is possible to use RI-CLPM with latent variables I already tested a CLPM with latent variables and I was wondering if I could use RI CLPM , I couldn't find anything about it Thanks
Can a ri-clpm be run with one set of continuous variables (5 time points, evenly spaced) and one set of dichotomous variables (same 5 time points, evenly spaced). The dichotomous variables are pass/fail on a developmental test and the continuous variables are hours of activity.
I have tried adapting Hamaker's syntax, using WLS and theta parameterization, but I can't get the model to converge and I fear I'm missing something. Is there any example syntax out there for dichotomous variables? Is it valid to run a model like this?
I was able to get my model to run by removing the within-person centered variables and measurement error constraints for the dichotomous variables. However, I think that changes the interpretation a bit, and that's where I'm struggling.
I look forward to an update about this ongoing research.