falsification test
A check that runs the analysis on data from before the policy took effect, where the true effect should be zero; if large 'effects' appear anyway, the chosen control group is invalid.
As defined in Jacob Vigdor on the Seattle Minimum Wage
Where it comes up · 1
Most important claims · 1 of 1Search all claims for “falsification test” →
- Standard border-discontinuity methods (comparing Seattle to its immediate suburbs, as in Card and Krueger 1994) failed falsification tests: running the method on pre-policy data produced large spurious effect estimates, revealing underlying trend differences that make the suburbs a poor control.Jacob Vigdor on the Seattle Minimum Wage