A different understanding of probability in a probabilistic population projection model and its outcomes
Christina Bohk, University of Rostock
Thomas Salzmann, University of Rostock
Population projections have a high relevance for several societal aspects, for instance for political, economical and social decisions. The progress from deterministic to probabilistic projection models was an important enhancement to capture the forecast uncertainty of the future evolution of the vital rates and at least of the future population size and its composition. But most of the prevalent probabilistic or stochastic approaches still capture the forecast uncertainty with some restrictions. In this paper we discuss our findings using different types of one probabilistic population projection model (PPPM). But in spite of following some different well known stochastic projection approaches we are going to introduce a novel approach (PPPM). The main characteristic of the PPPM is the meaning of probability. However, the PPPM does not generate probability for future vital rates via generating assumption sequences with stochastic processes that are implemented in cohort-component matrices. Rather, the PPPM is a very flexible projection model that takes exogenous assumption sequences with an allocated occurrence probability into the computation process. These exogenous assumption sequences are generated with no predetermined method, therefore they can be generated by e.g. simple or complex extrapolation methods, expert judgement, stochastic processes or a mixture of different methods. The higher the probability of occurrence of an assumption sequence, the more often it will be chosen by the random number generator in the progression of n trials. The outcomes – e.g. n future population sizes – range between the limits of several defined confidence intervals.
Presented in Poster Session 1