Sociology has long been concerned with explaining variation rather than identifying universal laws that apply across all contexts. Yet experimental research in sociology is often criticized for limited external validity and weak scalability. In this talk, I argue that these criticisms mischaracterize the potential of experiments for sociological theory-building. Rather than aspiring to universal generalization, experimental designs can be leveraged as powerful tools for middle-range theorizing—if variation is explicitly incorporated into research design. Drawing on three research programs, I illustrate this approach. A lost-wallet field experiment in New York City demonstrates how informed sampling can uncover contextual variation. Lab-in-the-field behavioral games among Ugandan farmers reveal systematic differences in public goods provision. Survey experiments on partisanship and social selection show how situational variation shapes decision-making. Together, these examples show how experiments can identify scope conditions and clarify when and where sociological theories apply. The talk reframes experimental research as a method for theorizing about variation itself, directly engaging the conference’s core concerns with explanation, prediction, and confirmation.