Impact evaluation methodologies

When and how to apply non-randomized methods of impact evaluation

December 15, 2009

Speaker: Dr. Akhter Ahmed, IFPRI

While randomization is clearly a useful tool for rigorous impact evaluation designs, it is sometimes not feasible from an ethical, political or logistical point of view. Therefore, non-randomized methods - such as reflexive evaluation, propensity score matching, difference-in-differences, instrumental variables, and regression discontinuity design -are often used to select appropriate counterfactuals in such situations. 

Dr. Ahmed from IFPRI presented at the Delhi seminar series the challenges of constructing counter factual using five non-randomized methods of impact assessment: reflexive evaluation, propensity score matching, difference-in-differences, instrumental variables, and regression discontinuity design. 

To download presentation, click here.

Scroll to Top