Modeling Spatial Diffusion: Reaction to Wellhofer

Wellhofer's adaptation of Klingman's suggestion that spatial diffusion be modeled as an autoregressive spatial series is questioned on two counts. First, this approach is inappropriate for most classes of diffusion models, including dendritic diffusion systems. Second, the procedure by which Wellhofer has ordered the observations in his Argentine illustration produces a database that bears almost no resemblance to the underlying pattern of causation he wishes to examine and actually creates an additional source of bias in the analysis.