110 likes | 232 Vues
Experiences in Democracy and Governance Field Experiments at USAID. Brian Scholl Research and Evaluation Fellow, USAID DCHA/DG Research Fellow, IZA-Bonn. EDGE History. 1990s – comparative “impact evaluations” 2001 – Six country case studies
E N D
Experiences in Democracy and Governance Field Experiments at USAID Brian Scholl Research and Evaluation Fellow, USAID DCHA/DG Research Fellow, IZA-Bonn
EDGE History • 1990s – comparative “impact evaluations” • 2001 – Six country case studies • 2003 – Social Science Research Council recommendations • 2006 & 2008 – Finkel et al. cross-national, quantitative studies • 2008 – National Academy of Sciences (NAS) report • Goldstone, Garber, Gerring, Gibson, Seligson, Weinstein
NAS Recommendations • Pilot program of field experiments including use of RCTs. • Rebuild USAID’s institutional mechanisms for learning. • Develop better DG indicators at the sectoral level. • Conduct country case studies in clusters to place DG assistance in context of national and international factors.
Progress to Date • 6 Pilot projects underway in 8 countries • Expansion of fellowship programs to provide technical assistance • Methodology • Theory
Pilot Field Experiments • Indonesia – decentralization • Cambodia – local political processes • Dominican Republic – youth leadership • Merida – CBD with crime prevention focus • Peru – local government oversight • Ukraine – political party development • Non-Pilot field experiment projects
Pilot Projects: General Lessons Learned • Projects can easily get away from us – need to be on top of almost daily • Insertion point not obvious • Assumptions underlying many programs and “theory of change” need serious reconsideration
Ukraine • Political party strengthening program • Anecdotal evidence suggestive of an effect • Trying to emphasize management implications rather than “do a RCT” • Complex dynamic for violation of SUTVA • Behavioral experiment
Peru • Strengthen civil society’s local government oversight function • Principal treatment intervention: information and training • Initially challenging to get implementer to focus on one activity • Two parts: Overall effect (small N) and effect of one activity (large N) • Selection: pre-qualification randomization • Outcomes: new set of measures
Merida: Caribbean Basin Security Initiative • El Salvador, Guatemala, Panama • Classical CBD/CDD with “crime” focus • Treatment unit: neighborhood • Selection: prequalification then randomization • Outcomes data: Augmented Lapop survey with over-sampling
Going Forward • More Field Experiments • Expanding Post-doc Fellowship programs • More attention to research and theories of change • EDGE Progression (my view): • Stage 1: Proof of concept (drawing to a close) • Stage 2: Rigorous theoretical underpinnings (underway) • Stage 3: Selective project Selection (now) • Stage 4: Targeted study areas (Starting in next few months)
Research Opportunities • For fellows • Access to non-public data and projects • Strong movement to bring both technical expertise and projects “in-house” • Research and travel funds • Next round of projects: selective on targeted study areas • Very challenging problems: much room to contribute to methods or theory as byproduct of field experiments • Innovation lab