Mplus Discussion >> Significance Testing of Variance Components

Topics
Last Day
Last 3 Days
Last Week
Tree View

Edit Profile


Significance Testing of Variance Comp...

Mplus Discussion > Growth Modeling of Longitudinal Data >

Message/Author

Igor Himelfarb posted on Wednesday, August 07, 2013 - 2:02 pm

Hi--

What distribution does Mplus use to test the significance of the variance of the latent slopes and interceptn in growth models?

Thanks,

Igor

Bengt O. Muthen posted on Wednesday, August 07, 2013 - 2:36 pm

What's printed in the output for a single variance is simply the ratio of estimate to SE and the p-value is computed assuming that this has a normal distribution; this is the standard output. It is well known that this is not a great way to test variances.

Igor Himelfarb posted on Wednesday, August 07, 2013 - 4:46 pm

Thanks for the response. Any suggestions for a more sophisticated way to test the variances in Mplus?

Thanks,
Igor

Bengt O. Muthen posted on Thursday, August 08, 2013 - 3:02 pm

There are interesting references on this, but new methods have not yet been implemented in Mplus. See, e.g., the 2004 JRSS-B article by Crainiceanu-Ruppert and the 2001 Berkhof-Snijders article in JEBS.

Igor Himelfarb posted on Friday, August 09, 2013 - 11:59 am

Thanks a lot for the references!