Conducting Experiments in Multiple Contexts
Graeme Blair, Gwyneth McClendon
Abstract
In an effort to assess the generalizability of treatment effects across contexts, scholars (or teams of scholars) are increasingly conducting experiments around the same research questions in multiple country and subnational contexts. In this chapter, we categorize recent and ongoing efforts to conduct cross-context experiments into three types: "uncoordinated," "coordinated, sequential," and "coordinated, simultaneous." We discuss some practical trade-offs across these types, arguing that coordinated cross-context designs offer the most promise for meta-analyses. We then draw attention to four areas in which the current approaches arguably all fall short in facilitating cumulative learning about treatment effects and treatment effect heterogeneity across contexts. We conclude by proposing some ways forward to continue improving our approach to learning about generalizability across contexts.