CPS workers' decisions in cases of domestic violence: a Classification and Regression Tree Analysis

The overall aim of my research project is to analyzing the decision-making process in the Child Protection System (CPS). More specifically, the study focuses on situations where partner/adult violence is an issue, and the... [ view full abstract ]

The overall aim of my research project is to analyzing the decision-making process in the Child Protection System (CPS). More specifically, the study focuses on situations where partner/adult violence is an issue, and the rights of different stakeholders are competing. The mandate of professionals from the legislation is to protect the best interest of the child; however, they have often been criticized for their tendency to hold abused mothers accountable and to take custody away from them, if they cannot manage to protect their children from exposure to domestic violence. Despite the heated debate, few studies have been produced to support these claims.
The present study aims to provide evidence to this debate, in the attempt to better understand the paths of DV-indicated cases in the Child Protection System and the decisions made by professionals in these situations.
The analysis is carried out on a sample of 34000 cases reported to six CPS agencies in Ontario from 2008 to 2010, whose paths are followed for one year through the Ontario Child Abuse and Neglect Data System (OCANDS).
Bivariate and multivariate analysis is performed to explore both workers dispositions and case outcomes. In particular, Classification and Regression Tree Analysis (CART) (Brieman et al., 1984) is used to model the decision to open a case for ongoing services or to intervene with out-of-home placement. CART is considered the more appropriate technique, given the research objectives. In contrast to logistic regression, it allows for different predictors to come into play for each subgroup, showing complex interactions. In addition, its hierarchical tree structure better mimics human decision making and it is relatively easy to interpret for professionals, in order to improve services to clients and maximize resources.