What predicts legislative success of early care and education policies?: Applications of machine learning and Natural Language Processing in a cross-state early childhood policy analysis

Following the pioneering efforts of a federal Head Start program, U.S. state policymakers have rapidly expanded access to Early Care and Education (ECE) programs with strong bipartisan support. Within the past decade the enrollment of 4 year-olds has roughly doubled in state-funded preschool. Despite these public investments, the content and priorities of early childhood legislation–enacted and failed–have rarely been examined. This study integrates perspectives from public policy, political science, developmental science, and machine learning in examining state ECE bills in identifying key factors associated with legislative success. Drawing from the Early Care and Education Bill Tracking Database, we employed Latent Dirichlet Allocation (LDA), a statistical topic identification model, to examine 2,396 ECE bills across the 50 U.S. states during the 2015-2018. First, a six-topic solution demonstrated the strongest fit theoretically and empirically suggesting two meta policy priorities: ‘ECE finance’ and ‘ECE services’. ‘ECE finance’ comprised three dimensions: (1) Revenues, (2) Expenditures, and (3) Fiscal Governance. ‘ECE services’ also included three dimensions: (1) PreK, (2) Child Care, and (3) Health and Human Services (HHS). Further, we found that bills covering a higher proportion of HHS, Fiscal Governance, or Expenditures were more likely to pass into law relative to bills focusing largely on PreK, Child Care, and Revenues. Additionally, legislative effectiveness of the bill’s primary sponsor was a strong predictor of legislative success, and further moderated the relation between bill content and passage. Highly effective legislators who had previously passed five or more bills had an extremely high probability of introducing a legislation that successfully passed regardless of topic. Legislation with expenditures as policy priorities benefitted the most from having an effective legislator. We conclude with a discussion of the empirical findings within the broader context of early childhood policy literature and suggest implications for future research and policy.