Skip to content
ML Academy

Introduced on day 5

Difference-in-differences (DiD)

An effect estimate equal to the change in treated units minus the change in comparable untreated units over the same period.

In detail

It removes fixed differences between groups and common trends. Its key assumption is parallel trends: without treatment, both groups would have changed alike. Check pre-period trends and look for other simultaneous changes.