Wim Beyers posted on Tuesday, May 28, 2019 - 12:44 am
In a 3-way cross-level interaction analysis, I want to know whether the effect of a within-level variable (X) on outcome (Y) is moderated by the interaction of two between-level variables (W & Z), while controlling for the main effects of W and Z on Y. With: - X is a dummy code dichotomous variable. - W is a latent variable composed of two observed variables (w1 and w2). - Z is an observed variable.
For this, I used this syntax : USEVARIABLES ARE id X Y w1 w2 cZ; MISSING ARE ALL (999); WITHIN = X; BETWEEN = cZ w1 w2 ; CLUSTER = id; DEFINE: cZ = Z - (4.104); ANALYSIS: TYPE = TWOLEVEL RANDOM GENERAL; ESTIMATOR = MLR; ALGORITHM=INTEGRATION; MODEL: %WITHIN% s1| Y ON X; %BETWEEN% W BY w1* w2; W@1; WZ | cZ XWITH W; Y ON W cZ; s1 (b1); s1 ON W (b2); s1 ON cZ (b3); s1 ON WZ (b23);
My questions are: (1) Is this syntax adequate for the purpose above?