In an attempt to improve the quality of educational research, the US Department of Education's Institute of Education Sciences has provided funding for 65 randomized controlled trials of educational interventions. We argue that this research methodology is more effective in providing guidance to extremely troubled schools about how to make some progress than guidance to schools trying to move from making some progress to becoming high-performance organizations. We also argue that the conventional view of medical research - discoveries made in specialized laboratories that are then tested using randomized control trials - is an inaccurate description of the sources of advances in medical practice. Moreover, this conventional view of the sources of advances in medical practice leads to incorrect inferences about how to improve educational research. We illustrate this argument using evidence from the history of medical research on the treatment of cystic fibrosis.