Marginal Revolution University on YouTube · Open on YouTube
Experimentation, causal thinking and metric design · Proficient · Lesson 61 of 99 · 11 min video, then about 3 hours of practice
Quasi-experiments: difference-in-differences and regression discontinuity
After this you can recognise when a natural experiment exists, apply difference-in-differences and explain its parallel-trends assumption, and describe regression discontinuity.
Why it matters
When you cannot randomise (a price change in one region, a policy that applies above a threshold), quasi-experimental designs can still estimate causal effects. Difference-in-differences compares changes over time between affected and unaffected groups and rests on the parallel-trends assumption, which you must check. Regression discontinuity uses a cut-off. The trap is reaching for these methods to rescue a question that just needs an experiment.
How to use this resource
Marginal Revolution University's clear introduction to difference-in-differences with an applied example.