Session: Worker Effects on Child Protection Outcomes: Three Case Studies (Society for Social Work and Research 30th Annual Conference Anniversary)

Although social work research has had a long and important impact on how child protection systems operate, the bulk of that research has emphasized child, family, and community effects on outcomes such as investigation rates, substantiation of investigations, placement into foster care, placement stability, and reentry. Relatively little attention has been paid to what might be called system effects. In the technical language of causal inference, when important variables are left out of a theoretical model that explains outcome variation, the omissions are treated as problems of model specification or omitted variable bias. In practical terms, leaving system effects out of child welfare research is tantamount to leaving schools out of models that explain student outcomes. The proposed symposium addresses this gap.

Worker effects are the focus of this symposium. Through the lens of three case studies, we explore how worker effects are conceptualized and studied. The first paper examines the time workers use when carrying out child protective service investigations. First, we break the work of CPS investigations into the underlying processes -- what do workers do and how long does it take on average? Then, after reading the regulatory requirements and holding focus groups with front-line workers (caseworkers and supervisors), we constructed a survey instrument that asks workers to report on the time they commit to the investigation components. The survey was given to all workers, statewide, so that it is possible to measure office-based differences in task completion times and the connection to outcomes.

We extend our analysis of time use to the analysis of workload. Typically, studies of caseload and workload are based on case characteristics. We add to that perspective by considering worker characteristics and how much time it takes to complete an investigation in service of the decisions that must be made. To pull these threads together, we assess the risk that an investigation will lead to placement or a service referral. Both decisions require more time before the work is done. We then use case histories to align case characteristics with the work that is added to the worker's workload when service and placement referrals are made. The results are then used to propose a workload allocation methodology.

The third paper turns to the question of disparity, with a specific focus on worker-level decisions. First, we construct a record of cases assigned to each caseworker. Assignments are longitudinal and reflect the case assignments over time. Using Poisson count models, we analyze disparity ratios at the work level after controlling for case, worker, community, and administrative effects. With this perspective, we can refine our understanding of similarly situated children to include children with similar backgrounds from similar communities, children served by the same administrative office, and by the same worker.

Taken together, the three studies highlight why it is important to add system effects to our child welfare studies. In this case, we treat workforce, time, and decision-making as factors that influence what we know about outcomes.