|
Message/Author |
|
|
Hello, I have complex data so I need to use MLR to obtain robust standard errors. I have also used the model constraints command to obtain standard error estimates associated with indirect effects. Can I assume that these SE estimates are also robust to the nonindependence caused by the complex sample? Thank you for your help. |
|
|
Yes, you can. |
|
Fabien Long posted on Tuesday, December 17, 2013 - 9:29 am
|
|
|
Dear Drs. Muthén, I am modelling a contrast effect (Big Fish Little Pond Effect) using group mean centering and model constraint: model: %within% Y on indACH (b_within); %between% Y on classACH (b_betwn); model constraint: new (BFLPE); BFLPE = b_betwn - b_within; Additionally, I am interested in whether there is a cross-level-interaction between average group achievement (classACH) and another level 1 variable. However, once I include the random slope "Beta1j | " in order to generate the cross level interaction, the parameter labels for the model constraint part are not recognized. My question is: Is there a way to combine cross-level interaction and model constraint? Also, using stdyx there are no standardized results for the model constraint "BFLPE". Is there any way to compute standardized coefficients for the New/Additional parameter? Thanks in advance! |
|
|
Q1. Yes. But you don't put the parameter labels after the betaj | .. statement on Within, but on Between for the mean, variance and regressions involving betaj. Q2. You have to express the variances you want to use and express the standardization in Model Constraint yourself. |
|
Back to top |
|
|