Open Conference Systems, CLADAG2023

Font Size: 
A Bayesian Spatio-Temporal Regression Approach for Confounding Adjustment
Carlo Zaccardi, Pasquale Valentini, Luigi Ippoliti

Last modified: 2023-06-15


For an accurate evaluation of the harmful impacts of pollution on human health, confounding variables must always be taken into account. Unfortunately, it oftentimes happens that some confounders might result unmeasured, hence, within a regression framework, the parameter that represents the exposure's effect might no longer be recoverable. In this paper, the unmeasured confounder is represented by a linear combination of basis functions, a technique that has been used in the spatial confounding literature, and that we expand to spatio-temporal designs. To reduce dimensionality and confounding bias, basis coefficients are assumed to have spike-and-slab priors.