Shouldn’t be differences in back door variables
Assumes random placement into treatment and control groups
- Regression discontinuity (ch 20)
Difference in Differences (ch 18)
- Changes over time for treatment outcome
- Changes over time for control outcome