Education researchers seek to understand what works, for whom, in what circumstances. Unfortunately, educational environments are complex and research itself is highly context dependent. Faced with these challenges, some have argued that qualitative methods should supplant quantitative methods such as randomized controlled trials (RCTs) and meta-analysis. I disagree. Good qualitative and mixed-methods research are complementary to, rather than exclusive of, quantitative methods. The complexity and challenges we face should not beguile us into ignoring methods that provide strong evidence. What, then, is the proper role for RCTs and meta-analysis in medical education? First, the choice of study design depends on the research question. RCTs and meta-analysis are appropriate for many, but not all, study goals. They have compelling strengths but also numerous limitations. Second, strong methods will not compensate for a pointless question. RCTs do not advance the science when they make confounded comparisons, or make comparison with no intervention. Third, clinical medicine now faces many of the same challenges we encounter in education. We can learn much from other fields about how to handle complexity in RCTs. Finally, no single study will definitively answer any research question. We need carefully planned, theory-building, programmatic research, reflecting a variety of paradigms and approaches, as we accumulate evidence to change the art and science of education.