Correlated residuals among first-orde... PreviousNext
Mplus Discussion > Confirmatory Factor Analysis >
Message/Author
 Richard E. Zinbarg posted on Friday, April 19, 2013 - 11:49 am
Hi,
We have a question about the higher-order model specified by the syntax below:
t1S by t1PA*(t1l1)
t1AD(t1l2)
t1ES(t1l3);
t3S by t3PA*(t3l1)
t3AD(t3l2)
t3ES(t3l3);
t5S by t5PA*(t5l1)
t5AD(t5l2)
t5ES(t5l3);
t7S by t7PA*(t7l1)
t7AD(t7l2)
t7ES(t7l3);
T by t1S
t3S
t5S
T7S;
T@1;
t3S on t1S (auto1);
t5S on t3S (auto2);
t7S on t5S (auto3);
Our question is whether the autoregressive statements at the end apply to the residuals of the first-order factors (T1S through T7S) or to the first-order factors themselves?
Thanks!
 Linda K. Muthen posted on Friday, April 19, 2013 - 1:00 pm
It is the factors themselves. See the FAQ on the website Regressing on Residuals if that is what you want.
 Paulo Alexandre Ferreira Martins posted on Wednesday, October 05, 2016 - 1:46 am
Hello,

Trying to improve my measurement model i added 1 cov residual, as the observed variables were semantically similar. However, the "warning message on PSI Matrix" appeared..
I compared the original model (without freeing residuals) but didn't find differences, such heywood cases, major variances, factor loading >.40 and so on...

ps: The 3 LV (factors) have very high correlations. Nonetheless, original model didn't show problems with this model.

Thank you.
 Bengt O. Muthen posted on Wednesday, October 05, 2016 - 3:57 pm
If you can't find the reason for the Psi warning such as a factor corr > 1 - send output to Support along with license number.
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: