Implementing Algorithmic Decision-Making Tools in Child Welfare Systems: Practitioner Perspectives on Use and Usefulness

Decisions to screen child maltreatment reports are often inaccurate and inconsistent, which has prompted jurisdictions to develop algorithmic decision-making tools to supplement workers’ judgments. However, the effectiveness of such innovations relies on successful adoption and consistent use by frontline users. Prior research has examined barriers to the adoption of decision-making tools in child welfare settings, but few studies have explored the implementation of algorithmic tools. This study described the use of such tools in practice and examined factors that influenced practitioners’ attitudes and behaviors as they integrated the tools into their work. A qualitative case study informed by the Technology Acceptance Model (TAM) and Consolidated Framework for Implementation Research (CFIR 2.0) was conducted regarding two county agencies implementing algorithmic tools for child welfare screening decisions. Data collection included document reviews, interviews with child welfare practitioners and leaders, and focus groups with child welfare and technology professionals. Participants disclosed key individual and contextual factors that impacted their perceptions of tool usefulness, including implementation processes, tool complexity, policy landscapes, internal communication structures, and staff role differences. Consideration of these factors must be incorporated into the future development and implementation of data-driven interventions to maximize their capacity to improve human services professionals’ decision-making.