When it comes to making decisions, the government is often flying blind. The public sector faces a major evidence deficit, and government agencies often lack the resources needed to design and evaluate new initiatives based on rapid advances in data science. The former director of the Office of Management and Budget estimates that less than $1 of every $100 of government spending relies on even the most basic evidence.

RegLab is working to develop the research methods needed to significantly expand the evidence base for good government. How can we improve the application of AI or machine learning models to real datasets? What are the best ways for researchers to evaluate government programs when the dominant method proposed by social scientists — randomized-controlled trials — are often practically or politically infeasible? How might we strengthen the ability of our institutions to benefit from statistical learning?

Given the scale of challenges the public sector must tackle, governments need the right tools and insights to engage in meaningful policy innovation. We seek to develop the methods needed to enable these broad policy changes.