The Case for Evidence in Government

Data alone can’t provide insight into the effectiveness of government programs—sophisticated analysis is vital.

Although the U.S. government presides over what collectively must be one of the world’s largest data repositories, its capacity to use that data to build citizen trust and make informed, evidence-based decisions is severely constrained. As explained in an enlightening report recently issued by the bipartisan Commission on Evidence-Based Policymaking (CEP), the mere existence of data is a necessary but not sufficient condition for creating empirical evidence to inform decisions throughout the full lifecycle of public programs—enactment, funding, operation, reform, termination.   

The digitization of many facets of various activities the government funds through its $4 trillion annual budget has resulted in a data explosion at federal agencies. But that data needs to be synthesized into actionable information to satisfy taxpayers’ demands for better results and greater transparency. The CEP report makes clear that much remains to be done to achieve that goal and provides a comprehensive plan to improve access to federal data, strengthen privacy protections and expand the public, private and academic research communities’ capacity to analyze data.

CEP provides an insightful list of recommendations such as establishing a National Secure Data Service to enable and leverage capabilities across government, addressing statutory impediments that obstruct smart data use, and streamlining processes used to grant researchers access to data. The report appropriately emphasizes strong privacy protections and advocates for comprehensive risk assessments for publicly released data and for the use of better technology and greater coordination across government. To prioritize efficient evidence building, CEP points out the need to coordinate statistical activities, evaluation and policy research within and between departments and across levels of government.

The report provides a critically needed blueprint for the development of a data analytics vision and strategy that is sorely lacking across government and within individual agencies, where the capacity to perform program performance assessments is uneven at best. CEP’s recommendations could not come at a better time given President Trump’s directive to agencies to identify ways to reorganize the executive branch—including program and agency eliminations—to improve efficiency, effectiveness and accountability. To realize that vision, agencies should be gathering statistically derived evidence, rather than basing decisions on gut instinct and intuition, to support their plans for re-imagining government.

Much has been made about the development of web-based portals (,,, providing citizens with direct access to government data. Those efforts are truly valuable and significantly enhance transparency and accountability by answering the what, where and when questions involving the expenditure of taxpayer dollars. The why and how well questions are more complex.

Transparency has its benefits but government data frequently requires sophisticated analysis to yield clear insights. While those websites do open up government for public inspection, there’s still a compelling need for heightened access to data by researchers to enable the discovery of actionable evidence. Such access is currently made difficult or impossible due to statutory data sharing prohibitions, an inability to access or combine datasets from multiple agencies, and a general reluctance by agencies to share their data.

With evidence, we can show program performance, describe what taxpayers get or don’t get for their money, and report on accomplishments and failures. That kind of information facilitates decision-making, enhances engagement and collaboration, and hopefully builds citizen trust—something that has been on a downward trajectory for 50 years. Surely there’s no omnipotent solution to reverse that trend overnight, but a sensible step in the right direction would be to strengthen government’s ability to use data to generate evidence in an efficient and technologically enabled way, with a citizen-centric focus on results.

This need for evidence evokes Sherlock Holmes’ impatient cry: “Data! Data! Data! I can't make bricks without clay.” That lament could have applied as well to federal agencies for much of the last half century as they tried to build evidence without the proper essential materials of accurate and reliable data along with the requisite analytical skills. Given tremendous advances in computational power over the past several decades, data collection has been facilitated to such an extent that agencies are now inundated with more data, clay, than they know what to do with. The CEP report provides a detailed instruction manual on how to fire up the kiln and not only turn that clay into bricks within agencies but it also shows how to add some mortar and connect those bricks to build trust-generating performance evidence structures across the entire U.S. government.

All this talk of bricks and the ongoing debate about a wall might lead one to ask if it’s true that we don’t need no education. To leverage government data and statistical capabilities to improve the lives of Americans, we actually do.

Douglas Criscitello is a senior lecturer and executive director of the MIT Golub Center for Finance and Policy at MIT Sloan School of Management