Obtaining linear predictions PreviousNext
Mplus Discussion > Latent Variable Mixture Modeling >
Message/Author
 Paul Spin posted on Friday, January 12, 2018 - 12:00 pm
Is there an MPLUS command for obtaining post-estimation linear predictions; that is something similar to Stata's [predict var, xb] command? Specifically looking to obtain linear scores from a Poisson model.

Thank you
 Tihomir Asparouhov posted on Friday, January 12, 2018 - 5:42 pm
Maybe this will work for you: take user's guide example 3.7 and add the command
plot: type is plot3;
Run the example and click on plots, estimated probabilities and sample proportions, plot estimated probabilities conditional on a set of covariates. Enter the values for the covariates. You will get a plot of P(Var=i). From there you can compute
E(Var)=sum P(Var=i)*i

This is just for one individual at a time. For categorical and continous we have individual residuals see web note 20 but not for counts. Alternatively you can specify the above formula in model constraints and get SE as well. Note that there is no linear score for this model - E(Var) is not linear.
Back to top
Add Your Message Here
Post:
Username: Posting Information:
This is a private posting area. Only registered users and moderators may post messages here.
Password:
Options: Enable HTML code in message
Automatically activate URLs in message
Action: