Alleviating confounding in spatio-temporal areal models with an application on crimes against women in India

Assessing associations between a response of interest and set covariates in spatial areal models is the leitmotiv ecological regression. However, presence spatially correlated random effects can mask or even bias estimates such due to confounding if they are not carefully handled. Though potentially harmful, issues have often been ignored practice leading wrong conclusions about underlying covariates. In spatio-temporal models, temporal dimension may emerge as new source confounding, problem be worse. this work, we propose two approaches deal with fixed by effects, while obtaining good model predictions. particular, restricted regression an apparently—though fact not—equivalent procedure using constraints proposed within both fully Bayes empirical approaches. The methods compared terms fixed-effect selection criteria. techniques used assess association dowry deaths certain socio-demographic districts Uttar Pradesh, India.