Speaker: Isa Marques
Title: Navigating Spatial Confounding: Understanding causes and proposing
mitigating approaches
Abstract: Spatial confounding is a fundamental issue in regression models for spatially indexed data. It arises because spatial random effects, included to approximate unmeasured spatial variation, are typically not independent of the spatially varying covariates in the model. This dependence can lead to significant bias in covariate effect estimates, limiting their interpretability. The problem of spatial confounding is complex, and in recent years, it has become the topic of extensive research. This talk introduces spatial confounding, its causes, and possible solutions. Using a spatial mixed model formulation, we bring mathematical clarity to the mechanisms of spatial confounding, broadcasting the relevance of spatial scale and smoothing. Based on this, in a Bayesian framework, we propose models to address spatial confounding with minimal tuning by the user.
Zoom Link here
Note: Seminars are free and open to the public. Reception to follow.
