See https://stat.ethz.ch/pipermail/r-sig-mixed-models/2011q2/006218.html Isn't this just what you get if declare some of the the fixed effects to be random effects in an ADNB RE model? Somone please enlighten me.