Scientific evidence is regularly invoked in defense of one classroom practice or another. Yet teachers, for the most part, ignore these studies. Why? First, teachers may view research as somewhat removed from the classroom, with further translation needed for the practice to be ready to implement in a live setting. Second, teachers may judge a practice to be classroom-ready in general but delay implementation because their particular students and setting seem significantly different from the research context. Third, teachers may resist trying something new for reasons unrelated to its effectiveness--because it seems excessively demanding, for example, or because it conflicts with deeply held values or beliefs about what works in the classroom. Finally, teachers may be unaware of the latest research because they only rarely read it. No matter the reason, it seems many teachers don't think education research is directly useful to them. The authors opine that these teachers have it right and that the problem lies with researchers, not teachers. The first three obstacles listed above--two concerning applicability of research and one concerning perceived constraints research puts on practice--are products of the methods researchers use. Research seems irrelevant to practitioners because it does not pose questions that address their needs. Teachers feel constrained by research because they feel pressured to use research-approved methods, and research creates clear winners and losers among practices that may be appropriate in some contexts but not others. The root of these issues lies in two standard features of most studies: how researchers choose control groups and their focus on finding statistically significant differences. The norm in education research is that, for a finding to be publishable, the outcomes of students receiving an intervention must be noticeably different from the outcomes of an otherwise similar "control" group that did not receive the intervention. To show that an intervention "works," the research must show that it makes a positive difference relative to the control. The authors ask if such comparisons are realistic, reasonable, or even helpful for teachers; and what researchers can do differently to make it so.