Count variables PreviousNext
Mplus Discussion > Structural Equation Modeling >
Message/Author
 Jost Reinecke  posted on Friday, December 12, 2014 - 3:43 am
If I have a regeression model with latent variables as predictors and a negative binomial variable as dependent variable the standardized solution of the regression coefficients should not be used because there is no variance of the dependent variable in a classical sense and therefore no residual variance and no Rsquare (the output shows always a value of 1.000 regarding the Rsquare of the negbin variable).

Is that right or can I make any substantial use of the standardized solution (despite of the continuous part of the model)?

The dispersion parameter is zero in the standardized solution, one could also assume from the output that the standardization makes use of the dispersion parameter estimated by the NegBin model.
 Bengt O. Muthen posted on Friday, December 12, 2014 - 6:08 pm
Right. I would standardize only with respect to the predictors, so multiply by the estimated SDs for the predictors.
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: