Spatial Stratified Heterogeneity
Spatial Stratified Heterogeneity (Geodetector) · Also known as: Geodetector, GeoDetector
Spatial Stratified Heterogeneity, commonly known as Geodetector, is a framework introduced by Jinfeng Wang and colleagues in 2010 for measuring and detecting spatial heterogeneity in data and identifying environmental risk factors. It quantifies the degree to which a given factor (variable) explains spatial variation in an outcome and is particularly valuable for environmental epidemiology, ecology, and geographical analysis where spatial non-stationarity is common.
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 Geodetector when analyzing spatial data where global regression assumptions (homogeneity, independence) are questionable. Ideal for environmental health studies (identifying disease risk factors that vary spatially), ecological studies (understanding how environmental gradients structure communities), climate and land-use impact studies, and any field where spatial non-stationarity is expected. It is particularly valuable when multiple factors may interact or when you want to compare which factor most strongly stratifies the outcome.
Strengths & limitations
- Directly measures the degree to which a variable explains spatial heterogeneity without assuming global linear relationships
- Does not assume constant regression coefficients across space (handles spatial non-stationarity)
- Allows comparison of multiple stratification variables to identify the most important spatial drivers
- Accounts for categorical and continuous variables; flexible stratification schemes
- Computationally efficient and does not require parametric assumptions about error distributions
- Particularly suited to complex spatial patterns common in ecology and environmental epidemiology
- Requires clear definition of strata; results depend on how the study area is partitioned
- Best for moderate sample sizes; very large samples may show statistical significance for trivial effect sizes
- Does not directly estimate effect sizes or magnitudes, only the power to explain spatial variation
- Assumes strata are well-defined; arbitrary or overlapping strata may yield unreliable results
- Does not account for spatial autocorrelation within strata (assumes independence)
Frequently asked
How do I choose strata for my variables?
Strata should be defined a priori based on subject-matter knowledge (e.g., ecological zones, administrative boundaries, or theoretically meaningful thresholds). Avoid post-hoc data-driven stratification, which inflates Type I error. Sensitivity analysis across plausible stratum definitions is recommended.
What does a high q-statistic mean?
High q (close to 1) indicates the stratification variable explains most of the spatial heterogeneity in the outcome. It does not prove causation, only that the variable is strongly associated with spatial variation. Interpret within the context of ecological or epidemiological mechanisms.
Can I use Geodetector with continuous variables?
Yes. Continuous variables must be stratified into categories (bins) before computing q. The choice of binning scheme affects results; conduct sensitivity analyses across reasonable thresholds to ensure robustness.
How do I test for interactions between two stratification variables?
Compute q for each variable alone and for their joint stratification. If q(A ∩ B) > q(A) + q(B), the variables interact synergistically. If q(A ∩ B) < q(A) + q(B) but greater than either alone, they interact antagonistically.
Is Geodetector better than spatial regression?
Neither is universally better. Geodetector is more flexible for non-stationary spatial patterns and does not assume linearity or global coefficients. Spatial regression provides effect estimates and inference under parametric assumptions. Use both to complement each other: Geodetector to identify important stratification factors, spatial regression to quantify effects.
Sources
- Wang, J. F., Li, X. H., Christakos, G., Liao, Y. L., Zhang, T., & Gu, X. (2010). Geographical detectors–based health risk assessment and its application in the neural tube defects study for the C–H plane. International Journal of Geographical Information Science, 24(1), 107–127. DOI: 10.1080/13658810802443457 ↗
- Wang, J. F., Zhang, T. L., & Fu, B. J. (2016). A measure of spatial stratified heterogeneity. Ecological Indicators, 67, 250–256. DOI: 10.1016/j.ecolind.2016.02.052 ↗
- Song, Y., Wu, P., Gilmore, D., Zhang, Q., Feng, Z., Wang, J., & Lou, L. (2020). Spatial autoregressive modelling of functional traits: a study case on gap-phase regeneration in a Chinese subtropical forest. Journal of Spatial Science, 65(2), 209–221. link ↗
How to cite this page
ScholarGate. (2026, June 3). Spatial Stratified Heterogeneity (Geodetector). ScholarGate. https://scholargate.app/en/sampling/spatial-stratified-heterogeneity
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.
- Cluster SamplingSurvey Methodology↔ compare
- Stratified SamplingSurvey Methodology↔ compare
- Systematic SamplingSurvey Methodology↔ compare