Legal Health Check: Can Predictive Modeling Create Tools for Legal Health Checks for Low-Income Communities

Background and Purpose: The purpose of this study is to understand the impact of various characteristic or experiences a person might have that effect the likelihood of being court involved and to assess legal risk. People become involved in Domestic Relations court and the impact of that involvement is often negative effecting mental, physical, economic health, and social well-being. Parents also face sometimes insurmountable barrier to accessing their children. This ongoing conflict between parent and the court system negatively impacts the parent-child relationship. Furthermore, there are few upstream approaches to preventing Domestic Relations court involvement even though prevention would produce more positive outcomes than crises interventions.

Methods: This study analyzed existing longitudinal data from The Fragile Families & Child Wellbeing Study conducted between 1998 and 2000 in large U.S cities. Data from Waves 1 of the study were used. Mother and or fathers who were interviewed at baseline were selected (N=956). The court involvement was measured by looking at two dependent variables MLegal_paternity and FLegal_paternity. These variables were nonbinary responses to the question of whether legal paternity has been established. There were five possible answers to this question. I don’t know, skipped, not in wave, yes and no. I don’t know, skipped and not in the wave were removed. We used 23 additional variables (characteristics) to build the prediction algorithm. 25% of the data was used to train the algorithm and the rest was used to test it.

Results: The training model had an accuracy rate of 100% by its last learning cycle after 5 learning cycles. Thus, the accuracy of the training model’s prediction was 100% by the last learning cycle. The training models’ error rate continued to decrease from .000 to .000 by the 5^th learning cycle. This means that the training model prediction error rate is .000%.

Implications: Although the research is limited, Domestic Relations court involvement has a negative impact on those involved. Preventing involvement in domestic relations court could improve individual and family well-being, especially the well-being of those who are low-income and or minorities. Understanding the characteristics and experiences that are likely to increase the risk of court involvement in the future will allow individuals and social service organizations to focus on those aspects of an individual or family and address them in such a way to prevent future court involvement. This training model is only the beginning of what a predictive legal health tool could look like. In addition to just predicting the likelihood of future court involvement the tool can provide information and legal resources to the individual based on the experiences or characteristics the individual has. Legal health is important to health and social work has played a limited role in promoting legal health and prevention. This is an important area for social work to address as it effects equity and justice.