The System for Using Evidence to Improve Federal Programs Has Been Gutted

Lauren Supplee recounts how three decades of building routine research, evaluation, and data use inside federal agencies transformed program design and management - examples include using research to set Head Start teacher wage rules, shifting welfare metrics from activities to employment outcomes, and informing child-care policy via the National Survey of Early Care and Education. That infrastructure was dramatically eroded by a recent reorganization that drove out thousands of scientists (estimates range from 10-40% of the federal scientific workforce), gutted research offices (Department of Education’s research arm lost about 90%, HHS research staff about 60%), rescinded permissions to publish, cancelled projects for political reasons, and sidelined career scientists. Supplee argues that statutory ambiguity - “may” versus “shall,” underfunded mandates like the Evidence-based Policymaking Act, and diffuse funding rules - enabled the dismantling while technically remaining in compliance.

She proposes rebuilding a resilient evidence-to-policy system with clearer laws and durable funding mechanisms: percentage-based evaluation set-asides tied to program growth and inflation, mandated evidence-building in grants, and funding for research-practice partnerships. She calls for legal protections for independent science, routine reporting to Congress on evidence creation and use, performance evaluations for leaders based on evidence use, and professional career tracks that translate research into timely decisions. Colleges should train practitioners for those roles; AI can assist but cannot replace trusted, skilled staff.