Theil index comparability across samples of different sizes

[Ignacio Pineda]

Hello,

I am currently using ineqdeco to compute Theil indexes on different income samples in order to use them in panel data analysis so I need those measures to be comparable across regions and time. Generalized Entropy measures do not range from 0 to 1 (like Gini indexes), but from 0 to a maximum of ln(N). I do not know if the ineqdeco command uses any type of normalisation that allows to compare indexes computed from samples of different sizes. If not, is there any way to make that normalisation?

Thank you!

[gokhan ekonomi]

Ignacio Pineda i have a smiliar question. i got some Theil (intra grooup-between group) results by using ineqdeco. if we assume that these samples reflect all population; can we compare these results with other stduies in which samples reflect populatiın also.

did u find any answer Ignacio?