Indiana University SPEA Edward J. Bloustein School of Planning and Public Policy University of Pennsylvania AIR American University

Roundtable: Using Administrative Data to Do Low-Cost RCTs
(Tools of Analysis: Methods, Data, Informatics and Research Design)

Thursday, November 12, 2015: 3:30 PM-5:00 PM
Pearson I (Hyatt Regency Miami)

*Names in bold indicate Presenter

Roundtable Organizers:  Erica Brown, Laura and John Arnold Foundation
Moderators:  Stuart Buck, Laura and John Arnold Foundation
Speakers:  Jon Baron, Coalition for Evidence-Based Policy, Benjamin L. Castleman, University of Virginia, Benjamin Goodman, Duke University and Mark Testa, University of North Carolina

A common misperception is that randomized controlled trials (RCTs) in the public sector are inherently expensive and cumbersome, requiring large contractors, multiple staff to implement new programs, collect data from scratch, and spend time following up on subjects by hand. Due to this misperception, many policymakers shy away from RCTs, thinking that it is infeasible to spend millions of dollars on new RCTs in a time of budget constraints. Yet RCTs need not cost millions of dollars and thousands of hours of staff time. In many instances, large, well-designed RCTs can be conducted at low cost and minimal burden by embedding random assignment in ongoing program operations and measuring outcomes using administrative data already collected by government or other entities for other purposes. These low-cost RCTs can build highly-credible evidence about whether programs are producing the hoped-for effects. In addition, low-cost RCTs can, for the first time, be used as a mechanism to test a variety of approaches to improving program impact, e.g., by bringing in insights from behavioral science about how programs can be more efficient and effective. This APPAM roundtable will build on a prior APPAM roundtable from three years ago by discussing new examples of large, low-cost RCTs that rely on administrative data to measure outcomes of high policy importance.