Level 2 variable as the outcome PreviousNext
Mplus Discussion > Multilevel Data/Complex Sample >
Message/Author
 Lixin Jiang posted on Monday, August 11, 2014 - 3:48 pm
I am testing a model where I have one level 1 predictor, one level 2 predictor, with a level 2 outcome. Is it possible to run MLM with random intercept and random slope with a level 2 outcome involving a cross-level interaction? Thanks.
 Bengt O. Muthen posted on Monday, August 11, 2014 - 5:19 pm
It sounds like you want a combination of UG ex 9.2 with its random slope and ex 9.4 with its level-2 outcome,which is possible.
 Lixin Jiang posted on Tuesday, August 12, 2014 - 8:03 am
The following is my code:

usevariables are id engag demanl2 stragl2;
!demanl2 is the level-2 predictor
!stragl2 is the level-2 outcome
!engag is the level-1 predictor

MISSING ARE ALL(-999);
cluster is id;
between are demanl2 dirl2;

analysis: type is twolevel random;

DEFINE:
CENTER engag(GRANDMEAN);

model:
%within%

s1| dirl2 on engag;

%between%
s1 dirl2 on demanl2;

I have this warning coming out:
*** ERROR in MODEL command
Between-level variables cannot be used on the within level.
Between-level variable used: DIRL2

Can you help me fix the issue? Thanks.
 Linda K. Muthen posted on Tuesday, August 12, 2014 - 12:06 pm
You put dirl2 on the BETWEEN list. You can't use it on within. dirl2 must be measured on the individual not the cluster level for a cross-level interaction. See Example 9.2.
 Lixin Jiang posted on Tuesday, August 12, 2014 - 12:17 pm
Thanks. Given that, the conclusion would be that Mplus doesn't allow a cross-level interaction with a level-2 outcome.
 Bengt O. Muthen posted on Tuesday, August 12, 2014 - 3:11 pm
I don't understand which model you are after. Perhaps you can write it out in formulas. A cross-level interaction implies that you have one exogeneous variable on level 1 and one exogenous variable on level 2 and their product influences the outcome which varies across both levels. This happens in a model with a random slope defined in a level-1 regression where the slope varies across the units on level 2.

A cross-level interaction is not a model where you have a random slope defined on level 2 - for that you would need a third level.
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: