Disentangling spatial interference and spatial confounding biases in causal inference.
Isqeel Ogunsola, Olatunji Johnson · 2026
Read Disentangling spatial interference and spatial confounding biases in causal inference. by Isqeel Ogunsola, Olatunji Johnson (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2602-02777.txt.
Description
Isqeel Ogunsola and 1 co-author pull apart two biases that corrupt causal estimates from spatial data, in work posted to arXiv in 2026. Spatial interference means one unit's treatment affects its neighbors. Spatial confounding means an unmeasured spatial pattern distorts the treatment effect, and its very definition is disputed in print.
Using directed acyclic graphs, the authors separate direct and indirect spatial confounding from interference by the bias each induces. Existing bias work assumes normal distributions for treatments and confounders, assumptions practice often breaks. The paper derives analytic bias expressions under more general distributions, using the Poisson case as an example. Spatial weights, the treatment's distribution, and interference size jointly determine interference bias. Weight matrices and exposure type govern indirect confounding. Model runs and a real spatial data set support the theory.
Catalog record
| CREATOR | Isqeel Ogunsola, Olatunji Johnson |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 85 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2602-02777