Dear Linda, I am running a mixture model with 5 latentt classes and within each class I fit a regression of the outcome on the explanatory variables (same model in each class). The aim is to see how the effect of explanatory variables differ from class to clas and hence I would like an output which puts Est./S.E for each explanatory variable in each class so that I can see which variable 'loads' on which class in a factor analytic sense. Is it possible to do that? Thanks
To do this you would need to use class-specific statements, such as
%c#1% y on x;
namer posted on Thursday, August 08, 2013 - 7:09 am
Dear Linda and Bengt,
I have a question regarding exporting FSCORES from a LCGA model. I am looking to export the means and standard errors of my latent intercepts and slopes from my 3 class solution into SPSS just to do some further exploration. However, using FSCORES the means are slightly different, however the standard errors are all very different, and nearly, if not all 0. I have read that factor scores are not the same as factors, and are calculated differently, so I expect this could be okay? My question is then, how do I export the estimates of the means and variances for individuals, so that group means and standard errors reflect what is reported in mplus output? Is this a possibility I have overlooked?