www.iita.org Research design and opportunities for evaluation USAID Feed the Future Africa RISING (Research in Sustainable Intensification for the Next Generation) Monitoring & Evaluation Expert Meeting, September 5-7, 2012, Addis Ababa, Ethiopia
Jan 28, 2016
www.iita.org
Research design and opportunities for evaluation
USAID Feed the Future Africa RISING (Research in Sustainable Intensification for the Next Generation)Monitoring & Evaluation Expert Meeting, September 5-7, 2012, Addis Ababa, Ethiopia
www.iita.org
Research questions
• What are options for SI poverty, nutrition, ecosystem stability?
• What are processes/mechanisms for integrating & delivering SI innovations?
• How improve livelihoods out of poverty/improve nutrition?
• How upscale/disseminate integrated innovations for SI beyond AR area?
www.iita.org
Motivation
• Need rigorous research design (hypotheses, parameters, puzzle, analytical methods, data collection) + appropriate counterfactuals/controls for confounding factors
• Need to move away from on-farm plot-level randomized trials to design that allows farmers to select into adoption
• Can use quasi-experimental and experimental (RCTs)
www.iita.org
Motivation
• Need global best practice: RCTs “gold standard”
• Need IPGs: internationally generalizable lessons: scalability
• Need to leverage public-private-civil society partnerships funded by USAID already on the ground
• Need to link supply and demand side experimentation
www.iita.org
Trials design: Random placement Trials design: Random placement
National Level
Regional Level
Agglomeration
Platforms: SI innovations
District
No intervention
“Counterfactual”
Obs. unit Village Village Village Village Village
TF1 TF2 No intervention
Limited/observation
Degree of intervention/ method
Limited/observation (RQ3)
Some/exploration (RQ2 and 3)
High/e.g., Fisher or classical experimentation: Mother-Baby; RCTS (RQ1)
Response variable dependent on the trials hypothesis
No intervention
Response variable dependent on the trials hypothesis
www.iita.org
Trials design: Random phase-in Trials design: Random phase-in
www.iita.org
ConsumersConsumers
Channel 2Fresh grains unprocesse
d
Channel 3Dried grains unprocessed
Channel 4Dried grains
export
ExportsExports
Channel 5Dried grains
processed
ConsumersConsumers
SubsistenceSubsistencehouseholdshouseholds
Consumption
Distribution
Trade
On-farm production
Inputs
Channel 1Subsistence production
& consumption
Processing
www.iita.org
• Use encouragement as an instrumental variable for adoption
• Random extension of credit to some farmers, randomized price subsidies through coupon distribution
• Randomized Mother-Baby trials
Trials design: Demand-side: Trials design: Demand-side: Encouragement Encouragement
www.iita.org
Trials design: Demand-side: Trials design: Demand-side: Encouragement: M-B Encouragement: M-B
www.iita.org
Trials design: Demand-side: Trials design: Demand-side: Encouragement : M-B Encouragement : M-B
www.iita.org
Trials design: Demand-side: Trials design: Demand-side: Encouragement : M-B Encouragement : M-B
www.iita.org
• Sell innovations at the market price
• Choose villages without innovations and randomly choose subset of villages to sell innovations
• Randomized offer of production and marketing contracts to entire treatment villages
Trials design: Supply-side: Introduce Trials design: Supply-side: Introduce to entire villages mimicking markets to entire villages mimicking markets
www.iita.org
Forward Contracts No Forward Contracts
Seed + Fertilizer Fertilizer Only Seed + Fertilizer Fertilizer Only
Seed Only Neither Seed Only Neither
Trials design: Supply-side: Introduce Trials design: Supply-side: Introduce to entire villages mimicking markets to entire villages mimicking markets
www.iita.org
• Offer some randomly selected groups of households in different settings at multiple sites the same technologies and delivery mechanisms
• GIS-based meta-analysis of the effects of the interventions
• GIS-based prediction of effects of subsequent implementation of the program
Scalability