Two level CFA Large residual variances PreviousNext
Mplus Discussion > Multilevel Data/Complex Sample >
Message/Author
 Souraya Mansour posted on Thursday, November 07, 2013 - 8:58 pm
I ran a two-level CFA model with 9 indicators, 3-factors at the child level and 2-factors at the classroom level. Should I be concerned about the unstandardized residual variances being too large particularly for 3 indicators: DELV, SI_PH and SI_GR?
What results in the residual variances to be so high? Are they acceptable to report?

Residual Variances
TOPELPK 92.482 6.495 14.240 0.000
TOPELPA 106.942 11.317 9.450 0.000
PALSNW 2.088 0.156 13.359 0.000
PPVT 66.939 7.272 9.205 0.000
NAPOLS 29.349 2.726 10.768 0.000
NAPCOMP 1.488 0.161 9.220 0.000
DELV 277.654 31.961 8.687 0.000
SI_PH 392.070 62.806 6.243 0.000
SI_GR 228.328 33.114 6.895 0.000

Thank you
 Bengt O. Muthen posted on Thursday, November 07, 2013 - 9:30 pm
The size of your residual variances is dependent on the scale of your variables. If the Sampstat output shows large variances for your variables there is nothing to worry about.
 Souraya Mansour posted on Friday, November 08, 2013 - 6:22 am
Great thank you!
One more question...when I add covariates to the model, the model does not converge (the covariates have a lot of missing data since they are supplementary surveys filled out by parents). What could be causing the problems with the model? Is it missing values (more than 50%)?
 Linda K. Muthen posted on Friday, November 08, 2013 - 2:58 pm
Please send the outputs, with and without covariates, 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: