From Design to Implementation Forging Partnerships and Mitigating
- Slides: 5
From Design to Implementation: Forging Partnerships and Mitigating Threats to Validity in Field Experiments Eric Mvukiyehe Columbia University Images: IRIN
Forging and Sustaining Partnerships What are possible outlets for potential partnerships? Government agencies Private institutions (incl. academic institutions) NGOs and international institutions Elites and political parties Takeaway: Different partners have different constraints & flexibility; find out about these to determine suitability for your project. What types of partnerships to negotiate? Partner-led collaborations Researcher-led collaborations Takeaway: Both types involve trade-offs & adjustment and certain conditions must be met for partnership to work.
Forging and Sustaining Partnerships… What are the conditions for collaboration? Potential for mutual interests: Feasibility of the experiment: Uncertainty about the outcome of an intervention; Intervention works, but mechanisms are not known; Low likelihood of harm to subjects and staff Flexible implementation (both practical and financial) Takeaway: Need to know when conditions are right for a field experiment. How to negotiate partnerships? Ability to play multiple roles, using different skill sets. Diplomatic skills Ethnographic skills Business consultancy skills Takeaway: Formalize agreements, esp. those pertaining to randomization, staffing and financing, data ownership and usage terms in MOU.
Threats to validity: causes and consequences Common threats to validity of field experiments No compliance Attrition (missing data) Interference & spillovers Takeaway: These problems can undo random assignment & complicate your ability to draw causal inference. Sources of threats to validity of field experiments Violation of randomization protocols (e. g. , replacing inaccessible communities) Poor administration of treatments (e. g. , esp. when several variation in treatments) Different objectives between researcher and partner Insufficient & untimely resources Takeaway: Need to foresee these problems and take preventive or mitigating measures.
Threats to validity: Mitigating measures Coordinate closely with partner on treatment assignment and administration, but maintain control over the research process Limit knowledge of the experiments =>reduce potential for compliance & spillovers Learn as much as possible about the research environment before and during implementation Ensure adequate resources & pay attention to logistics Documentation, Documentation Mistakes can only be corrected if they are discovered and documented Responsibility to report problems & mitigating measures in publications.