A Framework of High-Stakes Algorithmic Decision-Making for the Public Sector Developed through a Case Study of Child-Welfare

Algorithms have permeated throughout civil government and society, where they\nare being used to make high-stakes decisions about human lives. In this paper,\nwe first develop a cohesive framework of algorithmic decision-making adapted\nfor the public sector (ADMAPS) that reflects the complex socio-technical\ninteractions between \\textit{human discretion}, \\textit{bureaucratic\nprocesses}, and \\textit{algorithmic decision-making} by synthesizing disparate\nbodies of work in the fields of Human-Computer Interaction (HCI), Science and\nTechnology Studies (STS), and Public Administration (PA). We then applied the\nADMAPS framework to conduct a qualitative analysis of an in-depth, eight-month\nethnographic case study of the algorithms in daily use within a child-welfare\nagency that serves approximately 900 families and 1300 children in the\nmid-western United States. Overall, we found there is a need to focus on\nstrength-based algorithmic outcomes centered in social ecological frameworks.\nIn addition, algorithmic systems need to support existing bureaucratic\nprocesses and augment human discretion, rather than replace it. Finally,\ncollective buy-in in algorithmic systems requires trust in the target outcomes\nat both the practitioner and bureaucratic levels. As a result of our study, we\npropose guidelines for the design of high-stakes algorithmic decision-making\ntools in the child-welfare system, and more generally, in the public sector. We\nempirically validate the theoretically derived ADMAPS framework to demonstrate\nhow it can be useful for systematically making pragmatic decisions about the\ndesign of algorithms for the public sector.\n