Figure 9.

Residual spatial variation after adjusting the BYM model for an ecological covariate, the Cadum deprivation index (lung cancer mortality in the province of Lecce, 1992–2001). For males, MCMC estimate of residual spatial variation exp(v_i + ε_i) was reported for each area using a quintile based maps, in order to assess the presumptive effect due to the level of material deprivation in the i-th area. For females, the residual spatial variation considered to draw the map was exp(ε_i), as the clustering random effect did not enter the Bayesian model used for relative risk estimation.