Dear allstat group,
I have got three variables for longitudinal data (say) X, Y and Z, I
want to predict one of them on the other two variables, I mean to use
the three models for reporting the effects of any two on another. I am
using a Multilevel modelling or GEE technique for the analysis, the
three equation which I want to use are,
1 X=b0+b1Y+b2Z
2 Y=b0+b1X+b2Z and
3 Z=b0+b1X+b2Y
My Question is what statistical objection may be raised on this set up
and what would be a possible solution. the research question is I want
to predict X on Y and Z, Y on X and Z, and Z on X and Y?
|