Modeling Heteroscedasticity

[TinoNsenene1]

Dear all,

for a twolevel model, my research hypotheses suggest that not only the effect of a level-1 variable is moderated by a level-2 variable, but also the residuals of this crosslevel interaction term may be a function of a level-2 variable.

Stated differently, I am interested not only in modeling the expected mean of the level-1 random slope (which is the familiar cross-level interaction), but also its residual variance, or heteroscedasticity.

Could such a mean and dispersion model be fitted using the Mlwin software? Could this be done via the graphical user interface?

Best,
Tino

[ChrisCharlton]

It might be worth reading through chapter 7 of the MLwiN user guide (http://www.bristol.ac.uk/cmm/software/m ... al-web.pdf) to see whether this covers the kind of model that you are interested in, as this seems closest to what you describe.