FILE SELECTED // FILE 00
Difference in Differences
“It does not merely compare two groups; it compares how the gap between them changed.”
01 / SUMMARY
In short
The method subtracts the comparison group's time change from the treated group's change. Causal interpretation usually relies on plausible parallel trends and no relevant differential shocks.
03 / SCENE
In the wild
Sales rise 8 points where a rule changed and 3 where it did not; the estimated difference-in-differences is 5 points.
04 / LIMIT
The limit
Before-and-after data do not prove parallel trends or eliminate anticipation and composition effects.
FINAL NOTE
What remains
Inspect prior trajectories and concurrent events, not only the final subtraction.