jenny posted on Wednesday, March 30, 2005 - 9:20 am
Dear professor Muthen,
I have a problem concerning multilevel SEM: I have a dependent variable on both the individual and the group level. On the individual level, the variable is a latent construct with 5 indicators, on the group level, it is a single item. Can I model this in MPlus? I only found examples with dependent variables on the individual level until now.
You can have dependent variables on different levels. Observed variables can be declared to be either BETWEEN, WITHIN, or not declared such that they can vary on either level. These options should allow you to do what you want.
jenny posted on Tuesday, April 19, 2005 - 11:49 am
Thanks for your help. I started working with my model, but it didn't really work out so far. The problem is that I'm not sure how to relate my independent variables on the individual level to my dependent variable on the group level - is this possible at all? If yes, is there any example on this kind of multi-level model available from which I could learn how to do it?
Thank you very much in advance,
bmuthen posted on Tuesday, April 19, 2005 - 6:03 pm
In order for a variable observed for individuals (x, say) to influence a dependent variable on the group level (z, say), you have to have group-level variation in x (x = xb + xw, say) and let that part of x (xb) influence z. Here, x can be observed or latent. To work with such an x variable, you simply do not put x on either the within= or between= lists. If you have problems with getting this setup to work, please send input, output, data, and license number to email@example.com.