The past few years have seen a resurgence of faith in experimentation in education inquiry, and particularly in randomized controlled trials (RCTs). Proponents of such research have succeeded in bringing into common parlance the term "gold standard," which suggests that research emerging from any other design frame fails to achieve the rigor or significance of RCT-based research in answering causal questions and cannot reliably tell us "what works." In this article, Gary Thomas questions the reasoning behind this conclusion, resting his argument on the theory and practice of experimentation in education and on the limitations of RCTs in particular. He suggests that the arguments about the power of particular kinds of experiment reside in inappropriate ideas about generalization and induction and, indeed, what a scientific experiment needs to look like. Drawing from examples of systematic inquiry in education and other fields, Thomas argues for a restoration of respect for the heterogeneity of education inquiry.