Twolevel variances change according t... PreviousNext
Mplus Discussion > Structural Equation Modeling >
Message/Author
 Helplease posted on Monday, November 24, 2014 - 3:40 am
Hi,
I would really appreciate your help with a problem I've been having. I'm trying to do a Pitman's T test to compare two dependent groups (mothers and fathers). My data on these groups is on parenting and I have it for their twins, so that's two values for each parent. Since I didn't find a way in Mplus to do a test similar to the Pitman's T test (Is there a test for comparing variances of dependent samples?) I'm doing this outside of Mplus, but I still need the variances from Mplus in order to deal with the twins dependency as well. I'm therefore running a twolevel model with maternal behavior and paternal behavior as the only variables and I plan on using the between variances. The problem is that I noticed the variances change (for example from .178 and .142 to .179 and .128), if I add a third variable to the model (it also changes according to the third variable I use). Why is this happening?
 Linda K. Muthen posted on Monday, November 24, 2014 - 9:33 am
You can use MODEL TEST to obtain a Wald test. See the user's guide for more informaton.
 Helplease posted on Monday, November 24, 2014 - 10:31 am
Thank you Prof. Muthen!
Just to check that I understand correctly - the Wald test is suitable for comparing dependent samples(mothers/fathers)?
and as long as I'm using it under the twolevel model (under either the within or between option) it should cover the twins' dependency as well?
Also, could you please explain why the variances changed as a function of the variables in the model?

Many many thanks again!
 Linda K. Muthen posted on Monday, November 24, 2014 - 11:51 am
To have different coefficients for mother and father, you must have your data is wide format. The Wald test can be used to compare the coefficient of mother to the coefficient of father.
 Lior Abramson posted on Tuesday, November 25, 2014 - 1:26 am
Hello,
I also have a similar problem and would appreciate your help. I am running a two-level model with dependent groups, and I have noticed that if I add more variables to the analysis, I get different variances for the variables that were in the original model. Could you explain please why is this happening?
Thank you very much
 Linda K. Muthen posted on Tuesday, November 25, 2014 - 10:14 am
Please send two outputs that illustrate this issue and your license number to support@statmodel.com.
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: