You should probably enlist the aid of a professional statistician who can
provide the proper measure you need. The measure of diversity that I
mentioned is something you could try. Treat each variation of
religion/language type as a species and then calculate the relative
percentage of each species in the population. The diversity measure is
H= -Sum plnp for each species. If one species is 100% then H=0. H is
maximized if each species is equally distributed (50%/50% for 2 species).
You want the complementary measure, homogeneity rather than heterogeneity,
so you could take 1-H as your measure; it would be 1 if all members share
the same religion and language.
This would be an interesting study; perhaps it could shed some light on
genocide. Perhaps there is a critical range of heterogeneity where the
majority feels the need to reduce the heterogeneity. In societies that are
very homogeneous or very heterogeneous, you would not expect genocide.
|