Violence against children remains a serious social and public health problem in Indonesia, particularly in West Java Province, which consistently reports a high number of cases. Understanding the socioeconomic determinants that influence the spatial distribution of child violence is essential for developing effective and targeted policy interventions. This study aims to analyze the factors associated with the number of child violence cases across districts and cities in West Java using a spatial statistical approach. Specifically, the research examines the influence of several socioeconomic variables, including poverty rate, average years of schooling, divorce rate, Labor Force Participation Rate (LFPR), and Open Unemployment Rate (OUR), on the reported number of child violence cases.
The dataset consists of observations from 27 districts and cities in West Java. Preliminary statistical diagnostics were conducted to evaluate the suitability of regression models for count data. The results of these tests revealed the presence of overdispersion and spatial heterogeneity, indicating that traditional global models such as Poisson regression may not adequately capture the variability in the data. Consequently, the Geographically Weighted Negative Binomial Regression (GWNBR) model was employed, as it is capable of handling both overdispersed count data and spatially varying relationships between variables.
Model comparison was performed using the Akaike Information Criterion (AIC) to evaluate the performance of competing models. The results demonstrate that the GWNBR model provides a superior fit compared to both the Poisson regression model and the global Negative Binomial regression model. The GWNBR model achieved the lowest AIC value of 193.23, indicating its effectiveness in modeling spatially heterogeneous count data. The findings reveal that the influence of socioeconomic factors on child violence varies significantly across different regions. Among the predictors, average years of schooling and divorce rate were found to be statistically significant in the majority of districts and cities. In contrast, Bandung City emerged as the only area where all five explanatory variables significantly affected the number of reported child violence cases.
These results highlight the existence of geographically varying risk structures in child violence cases across West Java. Such spatial variability cannot be adequately captured by conventional global regression models, emphasizing the importance of spatially adaptive modeling techniques. The study provides empirical evidence that regional socioeconomic conditions play a crucial role in shaping patterns of child violence. Therefore, policymakers should consider local characteristics when designing prevention and intervention programs. Targeted and region-specific strategies are necessary to effectively address child violence and support the achievement of Sustainable Development Goals (SDG) 3, SDG 4, and SDG 16, which focus on health, education, and peaceful societies.
Author: Suliyanto, Drs., M.Si.
Details of research can be viewed here: https://www.researchgate.net/publication/400947749_Spatial_Analysis_of_Child_Violence_in_West_Java_Using_a_Geographically_Weighted_Negative_Binomial_Regression_Approach
