By studying the dynamics of small satellites around larger galaxies it is in fact possible to restrict the allowed parameter space for a set of mass models. This was achieved for the Andromeda galaxy by Geehan et al.  who analysed the stellar streams around M 31 and found a rather normal M / L for its disc and bulge. The original approach of Dehnen & King  included the study of the motions of 5 X-ray binaries (LMXB) detected in the Sculptor dwarf spheroidal. The rare occurrence of such binaries suggests that Sculptor must have retained all its LMXBs. Taking into account a visible mass of 107 M, and a measured stellar velocity dispersion of only 11 km s-1, the mass distribution of Sculptor should be dominated by a dark halo of 109 M within 1.5 kpc, which would imply a total M / L of at least a few hundreds. This tentative result is one of the very few constraints we have on the mass distribution of dwarfs, but we should realise that it will be a challenge to confirm the assumptions on which it is based.
I should add the mention of two recent illustrative works where stellar mass-to-light ratios of nearby dwarfs were derived via spectral line indices [42, 43]. Central counter-rotating core were revealed in both NGC 770  and VCC 510 , their spectral signature and stellar synthesis models indicating metal poor nuclei, the possible remnants of a merger/accretion event.
I will finally end up with a brief word on Modified Newton Dynamics (MOND). Apart from the fruitful debate it triggered 1, MOND is at least useful for one purpose: as emphasised by McGaugh , it proposes that we should view the mass discrepancy in the outer parts of galaxies as an empirical relation. This empirical relation is a very useful "tracer", which can then be analysed in the light of stellar population models, dark matter, MOND or any preferred prescription.
1 A personal opinion, obviously. Back.