CMStatistics 2022: Start Registration
View Submission - CMStatistics
B0335
Title: A Bayesian hierarchical time series model for estimating sex ratios in youth mortality Authors:  Fengqing Chao - King Abdullah University of Science and Technology (Saudi Arabia) [presenting]
Bruno Masquelier - Universite catholique de Louvain (Belgium)
Haavard Rue - KAUST (Saudi Arabia)
Hernando Ombao - King Abdullah University of Science and Technology (KAUST) (Saudi Arabia)
Leontine Alkema - University of Massachusetts Amherst (United States)
Abstract: Producing accurate estimates of sex ratios of mortality rates is essential in understanding population structure and dynamics, and in revealing sex discrimination. We introduce a Bayesian hierarchical model to estimate the disparity in age-specific mortality by sex for all countries over time, focusing on youth (15-24) mortality. The Bayesian model synthesizes data with varying levels, trends and associated uncertainties. The hierarchical modeling structure allows information exchange between data-saturated country periods and data-poor ones to assist estimation in country-periods lacking observations. Within the age groups 15-19 and 20-24, we model the global expected sex ratio using all the observations with a random walk model of order 2 (RW2). The RW2 is flexible to capture the non-linear global trend and is computationally efficient relative to splines model. We demonstrate the model-building process and motivate the choices of functions for the model elements. The model can be used to estimate sex disparity in mortality for other age groups, and we provide some illustrative results for modeling sex differences in 15-19 mortality. We conclude that the Bayesian model is an efficient and robust approach for estimating sex ratios of mortality.