Life table techniques for small populations: effects of population size, of the mortality level, and of age composition

Sergei Scherbov, International Institute for Applied Systems Analysis and Vienna Institute of Demography
Dalkhat M. Ediev, Vienna Institute of Demography

We study bias, standard errors and distributions of characteristics of life table with alternative ways of age grouping. Computations are based on the Chiang’s method, which was shown to have better statistical properties as compared to other methods (Eayres and Williams 2004). We generate events based on population structures of Russian regions and on several mortality schedules: actual Russian mortality schedule and model life tables. Simulations show that statistical efficiency of the methods is, first of all, affected by the population size. Yet, it may also be significantly affected by the population’s age composition. In order to facilitate practical choice of the rage of population sizes, for which life table computations may be considered as justifiable, we propose a simple theory for errors of classification of the mortality level in a given population. We propose a model and analytical solution for the probability of making an erroneous classification.