pl see the discussion in link

http://www.statmodel.com/discussion/messages/13/497.html?1484123077

ok.fine.contact  for further,if any.

I think the current consensus is to use the "R3STEP" command in Mplus.

See: Asparouhov, T., & Muthén, B. (2014). Auxiliary variables in mixture modeling: Three-step approaches using Mplus. Structural Equation Modeling: A Multidisciplinary Journal, 21(3), 329-341.

Hi Witold, this webnote may also be helpful for you as it includes Mplus syntax:

https://www.statmodel.com/examples/LTAwebnote.pdf

Kind regards, Karin

Hi Witold,

I am conducting now a LCA with a continuous covariate (PIPRE) using the new three-steps approach manually in MPlus. I managed to run it, but now I am not sure about how to interpret the output and then how to report the results.

As I read that you worked in something similar, I was wondering if you could give me some guidance or refer me to some article(s) where such results had been reported. I have only found articles with simulations studies.

My output looks like this:

Estimate S.E. Est./S.E. P-Value

Categorical Latent Variables

C#1 ON

PIPRE -0.244 0.420 -0.581 0.561

C#2 ON

PIPRE 1.442 0.627 2.301 0.021

Intercepts

C#1 0.684 0.268 2.556 0.011

C#2 -0.985 0.488 -2.018 0.044

LOGISTIC REGRESSION ODDS RATIO RESULTS

Categorical Latent Variables

C#1 ON

PIPRE 0.784

C#2 ON

PIPRE 4.229

ALTERNATIVE PARAMETERIZATIONS FOR THE CATEGORICAL LATENT VARIABLE REGRESSION

Parameterization using Reference Class 1

C#2 ON

PIPRE 1.686 0.535 3.150 0.002

C#3 ON

PIPRE 0.244 0.420 0.581 0.561

Intercepts

C#2 -1.670 0.461 -3.625 0.000

C#3 -0.684 0.268 -2.556 0.011

Parameterization using Reference Class 2

C#1 ON

PIPRE -1.686 0.535 -3.150 0.002

C#3 ON

PIPRE -1.442 0.627 -2.301 0.021

Intercepts

C#1 1.670 0.461 3.625 0.000

C#3 0.985 0.488 2.018 0.044

I would really appreciate some help.

Thanks!

Hi David,

Thank you very much for your quick response.

I am still not sure if I can interpret the results the way you suggest because they are log odds ratios and not probabilities. Shouldn't I first calculate the probabilities in some way in order to be able to make statements such as "Pipre is x times more likely to predict membership to class 2 than the reference class"?

Kind regards,

Macarena

Sorry for the repeated answers... browser turned crazy...

Thanks again. That's very helpful.

Have a nice week.

Best,

Macarena