Estimation of adult mortality using forward and backward projections.

Summary In this paper it is shown that, contrary to our intuitive understanding of the nature of population projection, the estimation of adult intercensal mortality leads to different results depending on whether forward or backward projection of the population is used. From this result a simple procedure is developed that yields estimates of the completeness of adult mortality registration. Finally, the nature and performance of a variety of methods that have recently been developed to estimate adult mortality in the absence of accurate data are compared.