Bayesian hierarchical models applied to subnational mortality estimation : three applications
(2023)
Files
Dissertation_bss.pdf
Open Access - Adobe PDF
- 17.3 MB
Details
- Authors
- Supervisors
- Masquelier, Bruno
- Abstract
- Subnational mortality estimation allows measuring spatial disparities in mortality and their evolution over time. It thus reflects a fundamental aspect of health inequalities. This however means estimating mortality indicators for small population sizes where the stochasticity in death counts is high. This, in turn, can lead to unclear underlying mortality levels. In these contexts, Bayesian Hierarchical Models (BHM) offer good performances, finding an appropriate balance between robustness and sensitivity. This dissertation is articulated around three applications where the research questions related to subnational mortality estimation are addressed thanks to statistical opportunities offered by BHM. First, I assess the possibility to estimate the probability of dying for children aged 5 to 14 years old at a subnational level in a sample of Sub-Saharan countries using survey data. Second, I measure the heterogeneity of the mortality shock in the context of the COVID-19 pandemic at the district level in Belgium. Third, I compare the performances of models allowing to estimate mortality age schedules at a subnational level. These three applications allow defining general guidelines for subnational mortality estimation.
- Affiliations
Citations
Schlüter, B.-S. (2023). Bayesian hierarchical models applied to subnational mortality estimation : three applications. https://hdl.handle.net/2078.5/100469