Estimation of exposure to toxic releases using spatial interaction modeling

Table 3 Regression results

OLS	no term	contain.	buffer	power	exp.	Cutter	RSEI
R-squared	0.4596	0.46	0.5224	0.4623	0.4596	0.5178	0.4599
Akaike Info. Criterion	22873.61	22873.22	22498.13	22860.47	22875.59	22527.45	22874.03
probability of TRI term	n/a	0.122677	<.000001	0.000104	0.877151	<.000001	0.209438
Spatial Lag
Akaike Info. Criterion	22875.6	22875.21	22350.12	22852.86	22876.37	22529.4	22876.02
probability of TRI term	n/a	0.91113	<.000001	0.96198	0.91903	0.83221	0.9158
Spatial Error
Akaike Info. Criterion	22873.95	22873.51	22298.14	22850.64	22874.72	22525.14	22874.57
probability of TRI term	n/a	0.19085	<.000001	0.13587	0.19759	0.038097	0.22659

Results for the multivariate ordinary least squares, spatial lag, and spatial error regressions of age-adjusted lung cancer mortality versus covariates and risk estimates from releases of lung carcinogens and related compounds calculated with each spatial interaction model. Bold entries indicate which spatial interaction model performed best. Note that lower values for the Akaike Information Criterion are preferred.

ISSN: 1476-072X