Spatial Analysis of Census Mail Response Rates: 1990–2010 - Space-Time Integration in Geography and GIScience - page 146

Fig. 9.2 Quantile choropleth maps: components of the random effects term. Increasing gray
scale is proportional to increasing values. To p ( a ): spatially structured random effects. Bottom
( b ): spatially unstructured random effects
negative spatial autocorrelation. Time lagged MRRs account for more than half
of the variability. Here the spatially structured component of the random effects
term comprises 38 positive spatial autocorrelation eigenvectors that account for
roughly 36.3 % of its variance. The resulting eigenvector spatial filter (Fig. 9.3 ),
whose MC SF is 1.01125 (indicating marked positive spatial autocorrelation;
MC SF /MC max D 1.01125/1.08718 D 0.93016), modestly reduces excess binomial