Improving Assessment of Severity of Violence in Social Service

Research problem: Custody cases often involve evaluation of the severity of physical and psychological violence occurring between the mother, father, and children.

However, how severe victims experience this type of violence may differ from how severe it is evaluated by the social worker(s) that evaluate the case.

This project aims to measure biases in the communication of severity ratings that victims are experiencing and the severity ratings of the social workers.

Previous research from our research group shows that several such biases are found in a normal population of intimate personal violence of victims and raters; including calibration, accuracy, and gender biases.

We hypothesize that these biases also are found in child custody disputes, but may be attenuated by education and computational methods.

Data and methods: The project will collect data from families with child custody disputes and social workers in Osby Social Services.

Parents will describe episodes of physical and psychological violence in a written format and rate the severity of the violence. These texts will be read and rated by social workers. The differences, or biases, between experience and evaluated violence will be measured.

Social workers will receive feedback on the differences in evaluations and subsequently be trained to attenuate identified biases.

The text will be analyzed by computational methods (natural language processes and machine learning) with the aim of providing a decision support system for the assessment of the severity of violence for social workers.

Societal relevance and utilization; Approximately 7000 child custody disputes and 3500 child protection services are taken care of by social services in Sweden annually, where correct evaluation of violence is crucial for the outcome.

The project aims at identifying and attenuating biases by educating the social workers and by introducing a decision support system.

Plan for project realization: The project applies for money to employ social workers to identify specific needs for social workers and collect data and a PhD student at Lund university to disseminate the results in scientific publications and create a decision support system. The project will primarily be developed in Osby social services and then disseminated in other social services.