Level-two variables with no complemen...
Message/Author
 Erich posted on Sunday, March 20, 2005 - 10:00 am
Dear all, if I understood the mplus manual for multilevel analysis correctly all examples of level-two variables have a complement variable on level-one. However, I wonder what happens when I have a level-two variable (e.g. % government spending on science education) with no complement for the within-groups matrix on level-one? Can I still use all the features of mplus for multilevel, in particular random slopes?

Erich
 Linda K. Muthen posted on Sunday, March 20, 2005 - 1:37 pm
The are BETWEEN and WITHIN options in the VARIABLE command to specify variables that exist on level 2 and level 1 only. So if you have a variable that is measured only on level 2 and therefore has no variation on level 1, you put it on the BETWEEN list. A random slope, however, must be defined on the WITHIN level using a variable that has variability on level 1.
 fritz posted on Monday, March 23, 2009 - 3:14 pm
Hello,
this is rather a beginners question. Is it possible to predict a between-level variable (without within-level variance) by both, between and within level predictors?

I tried the following using Mplus 5.2:
within = x1; between = x2 y;
model:
%within%
y on x1;
%between%
y on x2;

leading to the following error message:
*** ERROR in MODEL command
Between-level variables cannot be used on the within level.
Between-level variable used: y

I then tried not to specify y as a between variable resulting in error messages saying that y's variance approaches 0, which of course is true because it's a between-level variable. Thus, I'm not sure if this kind of model is appropriate at all.

 Linda K. Muthen posted on Monday, March 23, 2009 - 4:53 pm
Think what you want is as follows:

BETWEEN = z;
MODEL:
%WITHIN%
y;
%BETWEEN%
z ON y;
 Michael P. Marshal posted on Saturday, March 01, 2014 - 12:10 pm
Hello, I am asking current study participants to nominate peers for participation in a study of "hard-to-reach" or "hidden" youth. I'm anticipating that not all current participants will nominate peers thus some of my level-2 participants won't have level-1 units/data. Can i account for this dependency in mplus appropriately? And can it be handled in three level (trajectory) models? Thanks! Mike
 Linda K. Muthen posted on Sunday, March 02, 2014 - 5:08 pm
This would be handled by missing data.