A comparison of six analytical disease mapping techniques as applied to West Nile Virus in the coterminous United States

Table 2 Parameter estimation results for observed 2003 WNV death percentages.

	Gaussian SAR		Gaussian spatial filter		binomial spatial filter
Statistic	estimate	se	estimate	se	estimate	se
Spatial autocorrelation	0.08	0.101	1 (spatial filter MC = 0.243)	0.144	1 (spatial filter MC = 0.359)	0.090
intercept	-3.16	0.164	-3.15	0.105	-3.11	0.078
I₀	2.32	0.629	2.32	0.419	-18.21	25149
approximate dfs	45		39		40
Residual P(S-W)	0.029		0.212		0.522
Residual MC	0.0433		-0.0431 (z = -0.2)		-0.0927
Residual GR	1.0295		0.9720		0.9782
deviance	6.1357		4.2542		1.0198
pseudo-R²	0.425		0.643		0.292
predicted-observed regression slope	1.05		0.76		0.69
HGLM
	proper CAR				spatial filter
Statistic	estimate		se		estimate	se
spatial autocorrelation	0.09		0.104		0.999 (spatial filter MC = 0.359)	0.094
intercept	- 3.25		0.187		-3.12	0.072
I₀	-80.11		57		-78.86	60
approximate dfs	26				40
residual P(S-W)	0.018				0.001
residual MC	0.0119				0.0350
residual GR	0.7141				0.7494
deviance	0.4089				1.1660
pseudo-R²	0.697				0.442
predicted-observed regression slope	1.82				0.97

ISSN: 1476-072X