The Benefits of Randomized Experiments for Science and Society

Friday, February 15, 2013: 1:00 PM-2:30 PM
Room 206 (Hynes Convention Center)
Random experiments have allowed researchers to make advances in physical science, social science, industry, and health that have led to better policy and benefits to society. Randomized experiments use random assignment of study units to allow for learning when variation exists among the units and to estimate causal effects free of confounding by the selection of people into treatment conditions in social experiments. As a result, researchers now routinely test medical, educational, behavioral, and policy interventions with randomized experiments to identify effective interventions. This session provides examples of such uses and benefits of randomized experiments. The first paper describes applications of randomized experimentation to the study of politics, how experimentation has evolved in political science, and examples of its benefits to policy. The second paper describes examples of randomized evaluations of public health and education initiatives in developing countries. These experiments found that low-cost interventions could yield substantial health benefits, but paying teachers for performance had more mixed results. In behavioral health, treatment often follows a course of care, and determining the optimal course for a patient is challenging. The third paper describes how recent statistical advances have led to new experimental designs that allow the researcher to test for courses of care that yield optimal outcomes. It also gives an example of the use of such designs.
Daniel McCaffrey, RAND Corp.
Arthur Lupia, University of Michigan
Experimenting with Politics
Susan Murphy, University of Michigan
Experimenting To Improve Clinical Practice
See more of: Health and Pharmaceutical Science
See more of: Symposia