Error for multilevel analysis PreviousNext
Mplus Discussion > Structural Equation Modeling >
Message/Author
 Jean-Francois Bertholet posted on Tuesday, April 15, 2014 - 6:12 pm
I am from HEC Montreal university, where we start using Mplus.

I have a problem with a multilevel model. Can you please look at the error below with my syntax? Thanks!

variable:
names = A1-A231;
missing = all(-99);
usevariables are A193 A194 A195 A196 A197 A198 A199 A202 A203
A204 A205 A228 A229;
BETWEEN ARE A193 A194 A195 A196 A197 A198 A199 A202 A203
A204 A205;

cluster = A1;

DEFINE: CENTER A193 A194 A195 A196 A197 A198 A199 A202
A203 A204 A205 A228 A229 (GRANDMEAN);

analysis: type = twolevel;
model:
%WITHIN% ! Model for Within effects follows
LMX by A228 A229; ! LMX is a factor defined by A228 and A229
%BETWEEN%
HUMSUPP by A193 A195 A197 A198 A202 A204 A205;
HUMSUPN by A194 A196 A199 A203;
LMX on HUMSUPN;

output: sampstat stdyx ;

* ERROR
One or more between-level variables have variation within a cluster for
one or more clusters. Check your data and format statement.

Between Cluster ID with variation in this variable
Variable (only one cluster ID will be listed)

A193 3
A194 3
A195 3
A196 3
A197 3
A198 3
A199 3
A202 3
A203 3
A204 3
A205 3
 Bengt O. Muthen posted on Tuesday, April 15, 2014 - 7:05 pm
Look at cluster 3 and see if the variables A193-A205 have variation within that cluster. If they do, they are not Between variables as you have specified.
 Jean-Francois Bertholet posted on Tuesday, April 15, 2014 - 7:13 pm
Thanks for your rapid answer.

Maybe my understanding isn't clear, but what means "having variation within a cluster? It means that, for example, in cluster 3, all answers to A193 should be the same for all my respondants in Cluster 3?
 Bengt O. Muthen posted on Wednesday, April 16, 2014 - 12:17 pm
That's right. If the variable values are not the same for all respondents in the cluster, then the variable is not a between variable, that is, it varies not only across clusters but also within.
 Jean-Francois Bertholet posted on Friday, April 18, 2014 - 7:05 am
Thanks. I tried and the error was gone. One thing i don't understand now. How can i test a level 2 variable impact on a level 1 variable. Exemple, Supervisor Support at Group Level impact on Satisfaction at individual level. I seem to be able to test on one level, Mplus ask me to chose one or another? If you look this syntax, i want to test Lmxb on HUMSUPN.

analysis: type = twolevel;
%WITHIN% ! Model for Within effects follows
LMXw by A228 A229; ! LMX is a factor defined by A228 and A229
%BETWEEN% ! Model for Within effects follows
LMXb by A228 A229; ! LMX is a factor defined by A228 and A229
%WITHIN%
HUMSUPN by A194 A196 A199 A203;
%BETWEEN% ! Model for Between effects follows
HUMSUPNb by A194 A196 A199 A203;

%BETWEEN%
LMXb on HUMSUPNw;

output: sampstat stdyx ;
 Linda K. Muthen posted on Friday, April 18, 2014 - 9:00 am
See Example 9.2. You need a random slope for a cross-level interaction.
 Jean-Francois Bertholet posted on Wednesday, June 04, 2014 - 1:07 pm
Thanks.

Can you explain me what is the variable "w" in Example 9.2

Thanks for your time.
 Linda K. Muthen posted on Wednesday, June 04, 2014 - 2:42 pm
A between-level covariate.
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: