Abstract
The Structured Decision-Making Model’s risk assessment (RA) is a prominent feature of front-end child protection work. Examining how two different states have set policy to implement the RA, we analyze their distinct choices and unintended consequences. We propose that variation in implementation originates not from individual workers but from two nested sources. First, the RA embeds an implicit epistemological conflict between actuarially based and clinical decision making, with very little guidance on how to reconcile these different approaches into its design. Second, without explicit guidance on how to address divergence between scores and clinical judgment, states are free to set different policies, which, in turn, creates the conditions for variation in implementation. Examining these relationships advances our understanding of the conditions under which the RA is able to achieve a standardizing function.