fracglm betareg fractional data mixed multilevel models

[Don Vicendese]

Hi All,

Does anyone know of a method to do a regression on fractional data but in a multilevel setting?

Thanks,

Don

[Joseph Coveney]

If you're interested in just fitting a regression model that takes into account a clustering of related observations, then you could use something like

xtset id
xtgee fra . . ., family(binomial) link(logit)

which accepts fractional outcome values (proportions). It fits a so-called population average model.

Otherwise, you could look into using a beta density function (-help density_functions-) with -bayesmh- or -menl-.

[Don Vicendese]

Thanks Joseph. I ended up using fmlogit where you can have

fmlogit outcome_proportions, ///
eta(independent.variables) cluster(cluster.var) difficult iter(20)