|
![](http://www.statmodel.com/discussion/icons/csc/1px.gif) |
Variance in growth mixture model |
![Previous](http://www.statmodel.com/discussion/icons/csc/navgen_arrow_left.gif) ![](http://www.statmodel.com/discussion/icons/csc/1px.gif)
|
Message/Author |
|
csulliva posted on Tuesday, December 15, 2009 - 7:23 am
|
|
|
Is there a means by which the variance within latent classes can be assessed relative to the variance between classes in GMM (via MPlus output or otherwise)? |
|
|
There is not a specific part in the output that gives this, but you can compute the between-class variance as a function of the model-estimated parameters. You get the estimated mean in each class from which you can compute the overall estimated mean as a weighed sum of class-specific means using class probabilities as weights. Then compute the sums of squares of the difference between the class-specific and overall means. These computations can be done by hand or in Model Constraint using New parameters. |
|
Back to top |
|
|