1 Background

The methodologies described in this paper belong to a small subset of a broader set of methods developed to produce adjusted estimates of adult mortality for countries in the Latin American and Caribbean (LAC) region covering 150-160 years, from 1850 to 2010. This period encompasses approximately the end of colonial rule, the aftermath of wars of independence from Spanish and Portuguese domination, the establishment of nation states, integration into a world system and the world economy, and all developments that unfolded following World War II. 1 In this paper we focus only on adjustments of life tables for the post-1950 period. To do so we avail ourselves of mortality data consisting of yearly deaths by age and gender and population censuses. Because methods to adjust for completeness of death registration are well-known we focus on the description of relatively new methods to adjust for adult age misreporting. We then combine these two methods in an evaluation study designed to identify an optimal strategy to construct adjusted life tables for adult ages.

The paper is in six sections. In the first section we briefly define problems caused by defective vital statistics and census enumerations. In the second section we propose a model to represent the nature of adult age misreporting and in the third section we describe a methodology to detect and adjust for adult age misreporting. The fourth section describes an evaluation study designed to assess the performance of techniques to correct mortality indicators for both errors of coverage and age reporting. The fifth section discusses results from the evaluation study. The last section summarizes results and argues that adjustment of imperfect mortality data is subject to uncertainty and that treatment of the adjusted data is best carried out with models that account for uncertainty.