Stephen Raudenbush and Veronica Wald will use data from existing trials of intervention programs to illustrate how the choices made by researchers about the design and sample influence what researchers can learn about variation in program effects. The investigators aim to develop new strategies and user-friendly software to analyze variation in program effects, how programs operate, why program effects are larger in some settings, and if they affect youth who have not participated in the intervention. They will apply the framework and tools to data from recent multi-site trials of programs that primarily serve youth ages 8-25. The data sets will include evaluations of interventions on after-school instruction, charter school lotteries, job training, and small secondary schools.
Why does a youth program work better in some places than others? What accounts for this variability?