. 2011 Jul 21;6(7):e22192. doi: 10.1371/journal.pone.0022192

Table 3. Parsimonious model of 2005 district-level JE incidence as a function of climate, agriculture and land-use.

Variable	β Coefficient^†	S.E.	95% Confidence Interval (β)		Probability
Intercept	0.292	0.084	0.128	0.457	<0.001
April- p	−0.032	0.009	−0.051	−0.014	<0.001
*Pig-human*	0.637	0.516	−0.374	1.648	0.217
*Pig-human²*	−6.053	3.856	−13.610	1.505	0.117
*Irrigated*	0.033	0.012	0.010	0.056	0.005
*Grassland*	−0.033	0.022	−0.077	0.010	0.134
ρ	0.689	0.086	0.520	0.859	<0.001
R²	0.8122
Adjusted R²	0.7986
AIC	−89.933

p = mean precipitation, Irrigated = natural log of percentage of irrigated land per district, Pig-human = pig-to-human ratio, Grassland = natural log of percentage grassland cover per district, Rho ( ρ ) = spatial lag parameter.

S.E. = Standard error of â coefficient.

^†

All regression β coefficients represent a non-linear increase (or decrease when the coefficient value is negative) in JE incidence when there is a 1-unit increase in each respective predictor variable.