Dear all, I am trying to specify a model with two crossed random effects (response are clustered in items and in persons), including a latent aggregation of the response variable. This is the MPlus Syntax that I want to use:
DATA: File is Mplus_RT.dat; VARIABLE: Names are ID item response RT_log; MISSING ARE ID item response RT_log (-999); USEVAR ARE ID item response RT_log; CLUSTER = ID item; ANALYSIS: Type is CROSS RANDOM ; MODEL:
%within% RT_log on response (b_within);
%between ID% RT_log on response(b_betwn);
new(bflpe); bflpe = b_betwn - b_within; OUTPUT: sampstat stand tech1;
However, I get the error that "the following variable cannot exist on more than one level: RESPONSE"
When I specify only the random effect for ID, but not item, and change type to TWOLEVEL, the model with latent aggregation runs smoothly. Is there any possibility to specify a latent aggregation of the response variable with crossed-random effects?