Print

Synopsis

Many people living in low- and middle-income countries are not covered by civil registration and vital statistics systems. Consequently, a wide variety of other types of data, including many household sample surveys, are used to estimate health and population indicators. This talk describes a method of combining data from sample surveys and demographic surveillance systems to produce small area estimates of child mortality through time. Small area estimates are necessary to understand geographical heterogeneity in health indicators when full-coverage vital statistics are not available. For this endeavor, spatio-temporal smoothing is beneficial to alleviate problems of data sparsity. Conventional hierarchical models are not immediately applicable since one must account for the survey weights in order to alleviate bias due to non-random sampling and non-response.

The application that motivated this work is estimation of child mortality rates in five-year time intervals in regions of Tanzania. Data come from Demographic and Health Surveys (DHS) conducted over the period 1980-2010 and two demographic surveillance system sites. In this talk, Dr. Wakefield will derive a variance estimator that accounts for the complex survey weighting and describe a simulation study that examines the properties of the estimator, with a comparison to a jackknife alternative. For the application, the hierarchical models considered include random effects for area, time, and survey, and models are compared using the conditional predictive ordinate (CPO). The proposed method is implemented via the fast and accurate integrated nested Laplace approximation method.

Bio

Jon Wakefield is Professor of Statistics and Biostatistics at the University of Washington. Previously he has held positions in the Department of Mathematics (1990-1996) and the Department of Epidemiology and Public Health (DEPH, 1996-1999) at Imperial College, London.  At DEPH he was part of the small area health statistics unit and worked on methods and applications in spatial epidemiology. More recently his research has moved from chronic disease spatial epidemiology to infectious diseases spatio-temporal modeling and small area estimation. He has a longstanding interest in Bayesian methods and computation and in 2013 published the Springer book Bayesian and Frequentist Regression Methods. He is a member of the Statistical Genetics Program Faculty and is an Affiliate Member in the Vaccine and Infectious Disease Division at the Fred Hutchinson Cancer Research Center.