Table 1. The result of multiple regression: |r| = MI^*b1+Z^*B.

	Window size of local Moran’s I (W)
	3×3	9×9	15×15	21×21	27×27	33×33	39×39	45×45	51×51	57×57
b1	0.764	0.125	0.245	0.318	0.395	0.474	0.543	0.608	0.679	0.757
se	1.421	0.342	0.205	0.158	0.136	0.125	0.122	0.125	0.135	0.151
t	0.538	0.365	1.199	2.014	2.907	3.778	4.443	4.851	5.044	5.028
p	0.591	0.716	0.231	0.044	0.004	1.64E-04	9.54E-06	1.36E-06	5.13E-07	5.55E-07
AIC	−5661.9	−5661.4	−5659.8	−5659.7	−5662.2	−5666.2	−5670.2	−5672.9	−5673.8*	−5673.1

^*The lowest AIC value.

r: topographical correlation (Pearson’s correlation of the voxel weights of the two simulated patterns; MI: global Moran’s I; b1: coefficient of multiple regression of avgMI; Z: random effects dummy variables for 300 volume-pairs; B: coefficient for random effects; se: standard error of b1; AIC: Akaike Information Criteria for the whole model fit.