Fig. 1From: Spatial smoothing in Bayesian models: a comparison of weights matrix specifications and their impact on inferenceUnnormalised weights for ‘area 43’, shaded black, in the Scottish lip cancer data set. Darker purple areas are neighbours with large weights, signifying a high degree of correlation with area 43, as defined by the weights matrix of the respective models, while white areas have weights very close or equal to zeroBack to article page