Xiao Zhang posted on Saturday, August 10, 2013 - 5:13 am
Hi Prof. Muthen. I'm running a latent class growth model with two covariates. I hypothesized that one covariate would predict the latent classes via the other covariate. This is the model that I used:
%OVERALL% i s | mathC3@-1 mathC4* mathC5* mathC6* mathC7@0; c#1 on phiSC1 msqSC1; c#2 on phiSC1 msqSC1; c#3 on phiSC1 msqSC1; c#4 on phiSC1 msqSC1; msqSC1 ON phiSC1; Model indirect: c#1 IND phiSC1; c#2 IND phiSC1; c#3 IND phiSC1; c#4 IND phiSC1;
%C#1% ...... ......
After running this model, I got the following error message: *** ERROR in MODEL INDIRECT command Statements in MODEL INDIRECT must include the keyword IND or VIA. No valid keyword specified.
What can I do? Is there any problem with my model? Thanks.
A categorical latent variable cannot be used n MODEL INDIRECT. You would need to do this using MODEL CONSTRAINT using the formulas in Bengt's causal effects paper on the website. An indirect effect is not a product in this cases.