David Hill posted on Saturday, May 31, 2008 - 6:23 pm
In a current project I am trying to model the effect of X on Z (dichotomous) as it operates through Y (ordinal).
The problem is that Y and Z are reciprocal, in that Z is a predictor of Y and Y is also a predictor of Z. In my paper, I am attempting to model the effect of X on Y and then effect of Y on Z.
This is a non-recursive model that I believe Mplus can handle. I only have the demo version with limited input in terms of variables. The models that I am estimating in each stage include over 10 control variables. The path part of the paper described here is the last section of the analysis. Can I estimate the full models in each stage (Y on X [and other independent variables] and Z on Y [and other independent variables] and then use Mplus to examine indirect relationship described above?
My main interest is in the calculating the indirect effect of X on Z through Y.