Premium
Bayesian geo‐additive modelling of childhood morbidity in Malawi
Author(s) -
Kandala NgiangaBakwin
Publication year - 2006
Publication title -
applied stochastic models in business and industry
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.413
H-Index - 40
eISSN - 1526-4025
pISSN - 1524-1904
DOI - 10.1002/asmb.624
Subject(s) - covariate , markov chain monte carlo , econometrics , bayesian probability , random effects model , small area estimation , categorical variable , homogeneity (statistics) , prior probability , bayesian inference , statistics , markov chain , mathematics , context (archaeology) , inference , computer science , geography , medicine , artificial intelligence , meta analysis , archaeology , estimator
This paper applies a geo‐additive generalized linear mixed model to describe the spatial variation in the prevalence of cough among children under 5 years of age using the 2000 Demographic and Health survey (DHS) data from Malawi. Of particular interest in the analysis were the small area effect of geographical locations (districts) where the child lives in the time of the survey and the effect of the metrical covariate (child's age) which was assumed to be nonlinear and estimated nonparametrically. The model included other categorical covariates in the usual parametric form. We assign appropriate priors, within a Bayesian context, for the geographical location, vector of the unknown nonlinear smooth functions and a further vector of fixed effect parameters. For example, the spatial effects were modelled via Bayesian prior specifications reflecting spatial heterogeneity globally and relative homogeneity among neighbouring districts, thus a Markov random field prior is assumed. Inference is fully Bayesian and uses recent Markov chain Monte Carlo techniques. Copyright © 2006 John Wiley & Sons, Ltd.