Spatial Placebo Test
Also known as: geographic placebo test, spatial falsification test, spatial robustness check, geographic spillover test
A spatial placebo test is a falsification check used in geographic or spatial causal-inference studies. The analyst applies the same estimation procedure to spatial units, boundaries, or zones where no treatment effect should exist — fake borders, shifted cutoffs, or buffer areas beyond spillover range — and checks whether a spurious effect emerges. A non-significant result in the placebo region supports the credibility of the main causal estimate.
Key highlights
- Provides direct, transparent evidence that the estimated effect is localized to the true treatment boundary and not an artifact of spatial confounding.
- Can be implemented using the same code as the main analysis, requiring minimal additional effort once the fake assignment is constructed.
- Permutation-based versions yield a full distribution of null effects, allowing exact inference without parametric assumptions.
- Detects spatial autocorrelation artifacts or unmodeled geographic trends that might otherwise pass unnoticed.
- Widely accepted by referees and editors as a standard robustness check in the spatial and quasi-experimental literature.
- Applicable across a broad range of spatial estimators: RDD, DiD, IV, matching, and synthetic control.
Intuition
This section is available to Pro members. Upgrade to Pro
How it works
This section is available to Pro members. Upgrade to Pro
When to use it
Use a spatial placebo test whenever you rely on a geographic or spatial identification strategy such as a boundary discontinuity, a spatial DiD, or a geographic instrument, and you need to validate that your effect is truly localized to the treated zone. It is especially important in studies using administrative borders as quasi-random cutoffs, or in policy-impact evaluations where spatial spillovers are a concern. Do not use it as a substitute for theoretical justification of your design; a passing placebo test increases credibility but does not prove causality. Avoid when the study area is so small or homogeneous that there is no meaningful placebo region to test.
Strengths & limitations
- Provides direct, transparent evidence that the estimated effect is localized to the true treatment boundary and not an artifact of spatial confounding.
- Can be implemented using the same code as the main analysis, requiring minimal additional effort once the fake assignment is constructed.
- Permutation-based versions yield a full distribution of null effects, allowing exact inference without parametric assumptions.
- Detects spatial autocorrelation artifacts or unmodeled geographic trends that might otherwise pass unnoticed.
- Widely accepted by referees and editors as a standard robustness check in the spatial and quasi-experimental literature.
- Applicable across a broad range of spatial estimators: RDD, DiD, IV, matching, and synthetic control.
- Does not provide a formal test of the parallel-trends or continuity assumptions; it is a supplementary check, not a replacement for assumption verification.
- The choice of placebo zone is subjective — selecting an implausible or too-distant placebo may give an uninformative null result that does not genuinely test identification.
- Spillover effects from the true treatment can contaminate the placebo region, making a near-zero placebo estimate difficult to interpret in settings with strong geographic spillovers.
- Permutation tests require many repeated estimations, which can be computationally intensive for complex spatial models.
- A passing placebo test can generate false assurance if the placebo region shares the same confounders as the treatment region.
Common pitfalls
This section is available to Pro members. Upgrade to Pro
Applications
This section is available to Pro members. Upgrade to Pro
Frequently asked
How far should I place the fake boundary from the real one?
It should be close enough to share similar background characteristics but far enough that treatment effects do not reach it. A common approach is to shift the boundary by the same bandwidth used in the main RDD or to use a buffer zone beyond the expected spillover range. Sensitivity checks across multiple distances strengthen the argument.
Does a significant placebo result automatically invalidate my main finding?
Not automatically, but it is a serious concern that demands investigation. A significant placebo may indicate spatial confounding, spillovers, or a misspecified model. You should explore whether the placebo zone shares a geographic feature correlated with both the fake boundary and the outcome before drawing conclusions.
How many placebo boundaries should I test?
For a permutation test, researchers typically use dozens to hundreds of randomly or systematically shifted boundaries to construct a null distribution. Even testing three to five alternative positions adds credibility compared to a single placebo check.
Is a spatial placebo test the same as a temporal placebo test?
No. A temporal placebo assigns a fake treatment date before the actual intervention to check for pre-trends. A spatial placebo assigns a fake treatment location to check for geographic confounding. The two address different threats to validity and can both be applied in the same study.
Can I use this test in non-RDD designs?
Yes. Any causal strategy with a geographic component can benefit from spatial placebo checks — including spatial DiD, geographic IV, and spatial matching — as long as a plausible placebo region can be defined where the treatment genuinely did not apply.
Sources
- 1.Buonanno, P., Montolio, D., & Vanin, P. (2009). Does Social Capital Reduce Crime? Journal of Law and Economics, 52(1), 145-170.
- 2.Dell, M. (2010). The Persistent Effects of Peru's Mining Mita. Econometrica, 78(6), 1863-1903.
You have read it. What now?
Cite this page
ScholarGate. (2026, June 3). Spatial Placebo Test. ScholarGate. https://scholargate.app/causal-inference/spatial-placebo-test