Statistical Analysis of Spatial Autocorrelation in Child Poverty Across Nigerian Local Government Areas

Child poverty in Nigeria exhibits strong geographic concentration, and formal spatial statistical analysis of its autocorrelation structure across local government areas provides evidence on poverty clustering mechanisms that non-spatial analyses cannot capture. This study analyses spatial autocorrelation in child poverty rates across Nigeria's 774 LGAs using the 2022 Multidimensional Child Poverty Index computed from MICS6 survey estimates. Global and local spatial autocorrelation were assessed using Moran's I and Local Indicators of Spatial Association. A first-order queen contiguity spatial weights matrix was constructed. Spatial lag and spatial error regression models were compared against OLS to determine the appropriate spatial model specification. Global Moran's I for child poverty rate was 0.61 (p < 0.001), indicating very strong spatial clustering. LISA analysis identified four statistically significant High-High poverty clusters concentrated in the North West and North East LGAs and Low-Low clusters in South West metropolitan LGAs. Spatial Breusch-Pagan test confirmed spatial dependence in OLS residuals (p < 0.001). The spatial lag model provided better fit than the spatial error model by AIC, with the spatial autoregressive coefficient rho = 0.54 (p < 0.001), confirming significant poverty spillovers across LGA boundaries. Poverty-reducing interventions in central cluster LGAs would generate spillover benefits in neighbouring LGAs. The study recommends spatially coordinated child poverty programmes targeting identified High-High clusters and neighbouring transition LGAs. Keywords: spatial autocorrelation, child poverty, LISA, Moran's I, spatial lag model