The classical dispersion theory gives you the (complex-valued) dielectric function [itex]\epsilon(\omega)[/itex] from a simple damped harmonic oscillator ansatz for the electrons in the medium interacting with the incoming em. wave.
Quantum-field theoretically you find very similar results as the retarded in-medium Green's function of the em. field. Quantum mechanical dispersion theory on this linear-response level is not so different from the classical theory.
Interesting. I am not that interested in complex mediums, just a simple multiplicative description.
You can't avoid it though. A dispersive media always implies a lossy media by virtue of the Kramers-Kronig relation. As vanhees states though, the simplest model for a dispersive media is a simple oscillator as modeled by the Debye relaxation (or its variants like the Cole-Cole, Cole-Davidson, etc.). I would use Debye relaxation as a basic start for modeling dielectric dispersion. As simple as it is, it's very useful in application as long as you realize that it is meant to be applied over a finite bandwidth.