Using family network data in child protection services.

Area of Science:

• Social Sciences
• Public Health
• Data Science

Background:

• Child abuse prevention is a critical global objective for social services.
• Accurate risk assessment is vital to prevent child maltreatment, but current methods have limitations.
• Evidence-based tools are used, but few incorporate social network data.

Purpose of the Study:

• To analyze the utility of social network data in enhancing child maltreatment risk prediction models.
• To evaluate the impact of integrating family network information with existing child protection data.
• To determine if family network data improves the accuracy of identifying children at risk.

Main Methods:

• Analysis of a New Zealand child protection services dataset containing approximately 5 million social relationships (1996-2016).
• Simulation of integrating this dataset with birth records to create comprehensive family network data.
• Comparison of predictive model performance using demographic data alone versus combined network and demographic data.

Main Results:

• Including family network data significantly improves the performance of child maltreatment risk prediction models.
• Models integrating birth records with network data demonstrated superior accuracy compared to those using only recorded data.
• Early access to integrated family network information at case notification markedly enhances predictive accuracy.

Conclusions:

• A child's family network is a crucial factor in predicting maltreatment risk.
• Linking commonly available datasets with child protection records can substantially improve risk assessment.
• Utilizing comprehensive, up-to-date family network information enhances decision-making in child protection services.