Table 1:

Percent absolute bias (95% CI coverage) of the following methods in estimation of τ^* in 200 simulations: matching with no bias correction (Match 1), matching with a GAM bias correction (Match 2), matching with correctly specified bias correction (Match 3), BART, Poisson regression with linear terms (PR 1), correctly specified Poisson regression (PR 2). ω refers to matching tolerances on pollutants, i.e. 0.15 means that matches are restricted to be within 0.15 standard deviations for each pollutant. S/N is the proportion of the sample retained after trimming, and τ^*/τ is the ratio of the true trimmed sample TEA to the true whole sample TEA.

	ω	S/N	τ*/τ	Match 1	Match 2	Match 3	BART	PR 1	PR 2
S-1 τ = 195011	0.15	0.51	0.07	0.42 (0.00)	0.01 (0.98)	0.01 (0.93)	0.05 (0.62)	0.22 (0.00)	0.01 (0.88)
	0.20	0.58	0.09	0.42 (0.00)	0.00 (0.98)	0.05 (0.46)	0.06 (0.57)	0.21 (0.00)	0.02 (0.86)
	0.25	0.63	0.11	0.38 (0.00)	0.04 (0.74)	0.09 (0.08)	0.06 (0.56)	0.20 (0.00)	0.02 (0.81)
S-2 τ = 186852	0.15	0.51	0.20	0.04 (1.00)	0.02 (0.81)	0.02 (0.8)	0.00 (0.90)	0.05 (0.12)	0.01 (0.87)
	0.20	0.58	0.25	0.03 (1.00)	0.03 (0.42)	0.03 (0.39)	0.00 (0.90)	0.06 (0.04)	0.02 (0.77)
	0.25	0.63	0.29	0.01 (0.94)	0.06 (0.08)	0.06 (0.06)	0.01 (0.92)	0.07 (0.00)	0.02 (0.80)
S-2 τ = 132818	0.15	0.51	0.41	0.29 (0.00)	0.04 (0.98)	0.03 (0.98)	0.03 (0.99)	0.04 (0.53)	-
	0.20	0.58	0.47	0.30 (0.00)	0.04 (1.00)	0.04 (1.00)	0.05 (0.97)	0.06 (0.34)	-
	0.25	0.63	0.53	0.29 (0.00)	0.04 (0.98)	0.03 (0.99)	0.07 (0.89)	0.07 (0.18)	-