Spatial Distribution of Malnutrition among Children Under Five in Nigeria: A Bayesian Quantile Regression Approach

Issues of malnutrition among young children in developing countries are gaining more attention of policy-makers because of the adverse effects on the well-being of people and economic of these nations. Anthropometric variables used for determining malnutrition are measured through z-scores where those whose measures fall into the extreme ends of the scores are considered malnourished. Conditional mean regression has been adopted to examine the determinants but often times, covariates would have effect on the mean, but have no substantial influence on more extreme quantiles. We adopt Bayesian quantile regression approach to measure the spatial distributions of childhood undernutrition at state and local government levels in Nigeria. Markov random fields and Bayesian P-splines were used as priors for the spatial and nonlinear components respectively and estimation was through MCMC technique. Results show the existence of north-south divide in undernutrition in Nigeria and that observed socioeconomic variables could have little influence on the distribution of undernutrition across space in the country.