Statistics, experimentation, causal reasoning · insight-1021
Date the treatment from the announcement and check controls for spillover before running a difference-in-differences
- Evidence type
- documented fact
- Difficulty
- medium
- Roles
- Data scientist, Data analyst
- Industries
- aviation, retail, telecom
Summary: situation, action, reason, boundary
Situation
A fare change on one route was announced three weeks before it took effect, and a competing route serves as the control.
Action
Set the event date at the earliest plausible behavioural response, model or drop the announcement window, plot an event study with leads, and ask whether control units are substitutes or share resources with treated ones. If so, use a more distant control or estimate the spillover.
Reason
Difference-in-differences assumes the treatment has no effect before it is implemented and does not affect control units.
Boundary
Announced policy, price and schedule changes with nearby comparison units.
Symptom or risk
Customers pull bookings forward before the effective date, contaminating the pre-period, and the control route absorbs diverted demand, so neither side is clean.
Exceptions
Unannounced changes on isolated markets.
Trade-offs
A distant control is less similar, which weakens parallel trends.
Pitfalls
Declaring anticipation absent because the pre-trend test passed.
Lessons and concepts
Claims and their sources
Difference-in-differences relies on a no-anticipation assumption, that treatment has no causal effect before implementation, and the standard potential-outcomes framework imposes SUTVA, which rules out spillover effects.