Expanded spatial Durbin model for analyzing stunting prevalence in Java Island

Abstract

Stunting is a chronic health problem that impacts children's physical and cognitive development, especially in developing countries like Indonesia. There are several exogenous factors that influence the prevalence rate of stunting. Therefore, when examining spatial aspects, there is a possibility that the spatial dependences are occured not only on the response variable but also on the exogenous variables. Thus, a model is required to consider these spatial dependences. The Expanded Spatial Durbin Model (ESDM) can be used to predict stunting prevalence influenced by exogenous variables. We used the Euclidean method to determine the inverse distance weight matrix, and the Moran Index test was applied to identify autocorrelation. By incorporating spatial effects and relevant exogenous variables, the model can provide more precise estimates of stunting prevalence in different regions. This modelling technique is very effective for increasing the accuracy of stunting prediction, considering exogenous factors such as malnutrition prevalence and human development index (HDI) where all variables contain spatial dependencies. For the study, we choose Java Island in Indonesia as one of regions with a highest population density and significant stunting rates. Using the proposed technique, we found that the malnutrition variable has much stronger effect on the stunting prevalance than the HDI variable.