Quick reference for impact evaluation and analysis methods commonly used in development research.
Compares changes over time between treatment and control groups. Requires parallel trends assumption.
When to use: Program rollout with baseline and endline data, staggered implementation across regions.
Matches treatment and comparison units based on observable characteristics. Reduces selection bias.
When to use: Non-experimental settings where randomization wasn't possible but rich baseline data exists.
Exploits sharp cutoffs in eligibility criteria to estimate causal effects near the threshold.
When to use: Programs with clear eligibility thresholds (income cutoff, test score, geographic boundary).
Uses an external source of variation to address endogeneity in treatment assignment.
When to use: When you have a valid instrument that affects treatment but not the outcome directly.
There was an error while loading. Please reload this page.