The purpose of the Design & Analysis of Experiments TIG is to provide an active forum for AEA members who specialize in or have special interest in experimental evaluation research to connect and advance the methods and practice of this segment of the evaluation field. Evaluations that use an experimental design (with randomization of treatment and control units) are distinctive and involve their own methods, practices and analyses, as well as many that connect to other TIGs’ foci. We hope to engage members from other relevant TIGs in order to have cross-fertilizing discussions.
Specific areas of scholarly discussion for this TIG include:
· integrating randomization into program operations
· how to analyze experimental data (ITT, ATE, TOT, LATE, CACE, moderator effects)
· design strategies for getting inside the black box
· analytic strategies for getting inside the black box
· opportunities to explore mediation with experimental data
· power analyses/considerations (e.g., re: differential effects, in clustered-designs)
· opportunities/challenges from multi-site experiments (when pooling is/is not justified; ICCs and power implications; extending multi-level modeling to experimental data)
· within study comparisons (under what conditions do varied quasi-experiments reproduce experiments’ results)
· handling missing data and attrition (when differential T-C attrition is a fierce enemy)
· the external validity of experimentally-designed evaluations
· low cost experiments
Consider becoming active in the TIG in order to shape its future and contribute to our field.