Spatial Propensity Score Weighting
Spatial Propensity Score Weighting for Causal Inference · Also known as: spatial PSW, geographically weighted propensity score weighting, spatial IPTW, spatially adjusted inverse probability weighting
Spatial propensity score weighting extends inverse probability of treatment weighting (IPTW) to settings where units are geographically located and treatment assignment may depend on spatial factors such as location, neighborhood characteristics, or spatial clustering. By incorporating spatial covariates into the propensity score model and adjusting standard errors for spatial autocorrelation, it produces more credible causal estimates from observational geographic data.
Read the full method
Sign in with a free account to read this section.
Method map
The neighbourhood of related methods — select a node to explore.
When to use it
Use spatial propensity score weighting when treatment assignment is geographically structured — for example, policies rolled out by region, environmental exposures that vary across space, or neighborhood-level interventions. It is appropriate when spatial covariates are available, units have geographic identifiers, and there is reason to believe that location-specific confounding or spatial spillovers are present. It is not appropriate when the data has no meaningful geographic structure, when the spatial sample is too sparse for reliable geographically weighted estimation, or when strong spatial interference (SUTVA violations) renders any weighting approach invalid without further design adjustments.
Strengths & limitations
- Explicitly accounts for spatial confounding by incorporating geographic covariates into propensity score estimation, reducing bias from location-based selection.
- Flexible: the spatial component can be added to standard logistic regression through spatial lag terms, geographically weighted regression, or area-level fixed effects.
- Produces a weighted pseudo-population that is balanced both on observed individual attributes and on geographic context, improving comparability.
- Compatible with robust spatial inference procedures (Conley standard errors, spatial cluster bootstrap) that account for residual autocorrelation.
- Applicable to cross-sectional and panel data with spatial identifiers, making it broadly usable in environmental, public health, and social science research.
- Requires spatial covariates that capture the relevant geographic confounders; if key spatial factors are unmeasured, the spatial propensity score remains incomplete.
- Geographically weighted propensity score models require sufficient local sample size; sparse data in some regions leads to unstable local estimates.
- Does not handle interference between units (spillovers) unless the estimand is explicitly redefined; standard IPTW estimands assume no interference.
- Spatial autocorrelation in residuals may persist even after weighting if the outcome model is misspecified.
- Extreme propensity score weights, which are a general IPTW problem, can be amplified when spatial confounders are strong, requiring trimming or stabilized weights.
Frequently asked
How does spatial propensity score weighting differ from standard IPTW?
Standard IPTW estimates the propensity score from individual covariates alone and ignores geography. Spatial PSW adds spatially referenced variables — location coordinates, area-level characteristics, spatial lag terms — to the propensity score model, and adjusts inference for spatial autocorrelation. This reduces bias from geographic confounding that standard IPTW would leave unaddressed.
What spatial covariates should I include in the propensity score model?
Include covariates that predict treatment assignment and are spatially structured: area-level demographics, distance to infrastructure, spatial lag of the outcome or treatment, or geographic coordinates (e.g., via spline terms). The goal is to absorb location-based differences in treatment probability that would otherwise confound the causal estimate.
How do I check whether weighting has removed spatial imbalance?
After computing weights, examine standardized mean differences for all covariates (including spatial ones) in the weighted sample. Additionally, map the geographic distribution of weighted treated and control units, and optionally compute a weighted version of Moran's I to test whether spatial clustering of treatment assignment has been reduced.
What standard error estimator should I use?
Use spatially robust standard errors such as Conley (1999) spatial HAC standard errors, which allow for correlation among units within a geographic bandwidth, or cluster-robust standard errors at a meaningful geographic unit (county, grid cell). Ordinary or heteroskedasticity-only robust SEs will understate uncertainty when residual spatial autocorrelation remains.
Can this method handle spatial spillovers between units?
Not directly. Standard spatial PSW inherits the stable unit treatment value assumption (SUTVA) from IPTW, meaning it assumes no interference between units. If spatial spillovers are substantial, the causal estimand must be redefined (e.g., direct and spillover effects), and more specialized methods for interference are needed.
Sources
- Keele, L., & Titiunik, R. (2015). Geographic Boundaries as Regression Discontinuities. Political Analysis, 23(1), 127-155. DOI: 10.1093/pan/mpu014 ↗
- Hirano, K., Imbens, G. W., & Ridder, G. (2003). Efficient Estimation of Average Treatment Effects Using the Estimated Propensity Score. Econometrica, 71(4), 1161-1189. DOI: 10.1111/1468-0262.00442 ↗
How to cite this page
ScholarGate. (2026, June 3). Spatial Propensity Score Weighting for Causal Inference. ScholarGate. https://scholargate.app/en/causal-inference/spatial-propensity-score-weighting
Which method?
Set this method beside its closest kin and read them side by side — the library lays the books on the table; the choice is yours.
- Geographically Weighted RegressionSpatial analysis↔ compare
- Inverse Probability WeightingCausal inference↔ compare
- Propensity Score WeightingCausal inference↔ compare
- Spatial Difference-in-DifferencesCausal inference↔ compare
- Spatial Propensity Score MatchingCausal inference↔ compare
- Spatial Regression Discontinuity DesignCausal inference↔ compare