Estimating latent class models in Stata using both categorical and continuous indicator variables

[James Lee]

Hi Statalist community,

I am trying to run latent class models and I am fairly new to this type of analysis. I have continuous and categorical indicators. I am using the following link as a reference. In the second example within the link, there is an example to run latent class models where there are both continuous and categorical indicators.

https://stats.oarc.ucla.edu/mplus/se...ixture-models/

However, the analysis is conducted using Mplus which is quite expensive and I am trying to replicate the example using Stata. I was wondering if there is a native Stata command or a user-generated Stata package available for me to replicate the example. In addition, if you have any papers that you could direct me to, I would really appreciate it. Thank you so much.

[William Lisowski]

Look in the Structural Equation Modeling Reference Manual PDF included in your Stata installation (assuming you have a relatively recent version) and accessible from Stata's Help menu. Note that examples 50g and 51g discuss using the gsem command for Latent Class modeling.

Having done that, understand that if you are unfamiliar with structural equation modeling, you need to start at the top to gain a basic understanding of how it is done in Stata. Latent Class modeling is just a specific application of the more general concepts.

[James Lee]

Hi @William Lisowski,

Thank you for your suggestion. I read the manual that you suggested. I used the link below.

https://www.stata.com/manuals/sem.pdf

I read the section in 50g for latent class analysis. The example includes indicators that are strictly dummy variables. On page 486, these variables are accident, play, insurance, and stock. I also read the section in 52g for latent profile analysis which includes indicators that are strictly continuous variables. On page 494, these variables are glucose, insulin, and sspg. However, I don't see a latent model example in the manual where there are both continuous and categorical indicators.

There was a similar question asked in an earlier Statalist post. The link is below. A suggestion was given but I don't see documentation or an example in the Stata manual to confirm that this is the correct approach. I want to be sure that the suggestion in the earlier Statalist post is accurate. Can someone confirm the suggested code is accurate? Thanks.


https://www.statalist.org/forums/for...ous-indicators