The Probabilistic Life Table and Its Applications

Nan Li, United Nations
Shripad Tuljapurkar, Stanford University

Life-table variables are assigned certain values at all ages, and hence are treated as deterministic. Moreover, the starting number of the hypothetical cohort is taken arbitrarily as 100 thousand. But the survival process of the hypothetic cohort is uncertain unless the starting number is infinite, and the uncertainty depends on the starting number. Thus, the nature of life tables is probabilistic, and the starting number of the hypothetic cohort should not be arbitrary. This paper provides a method to compute the probabilistic life table that uses a specific starting number for the hypothetical cohort. In a probabilistic life table, each variable has a probability distribution rather than a sample value, at all the ages except the first one. Using probabilistic life tables, one can tell, for example, whether the difference between two life expectancies is statistically significant or not. Finally, deterministic life tables are approximations for large populations.

  See paper

Presented in Session 122: Mathematical and Computational Models and Methods