Table 2. Negative binomial regression model^a for the rate of inactivation of E. coli O157:H7^b on heads of romaine lettuce (n = 559) over 10 days in central coastal California during July–Aug. and Oct., 2012.

Factor	Factor Categories	Coefficient	Std. Err.	95% C.I.	P-value
Fecal	Pigc	0.0
	Rabbit	−1.232	0.334	−1.888, −0.576	0.000
	Chicken	−1.289	0.342	−1.959, −0.619	0.000
Trial	Julyc	0.0
	October	−4.066	0.375	−4.801, −3.331	0.000
Trial × Fecal	July × Pigc	0.0
	October × Rabbit	1.929	0.487	0.974, 2.884	0.000
	October × Chicken	2.016	0.480	1.076, 2.956	0.000
Daily leaf wetnessd		0.041	0.012	0.017, 0.065	0.001
Sampling evente	1 (0.5 ± 0.3)c	0.0
	2 (21.8 ± 0.8)	−0.657	0.391	−1.423, 0.109	0.093
	3 (44.6 ± 0.1)	−1.565	0.490	−2.526, −0.605	0.001
	4 (68.9 ± 0.4)	−4.170	0.408	−4.97, −3.371	0.000
	5 (93.2 ± 0.5)	−5.642	0.408	−6.441, −4.842	0.000
	6 (115.9 ± 0.1)	−3.880	0.450	−4.762, −2.998	0.000
	7 (119.5 Oct.)f	−6.167	0.517	−7.179, −5.154	0.000
	7 (123.0 July)f	−0.133	0.655	−1.417, 1.150	0.839
	8 (141.7 ± 0.5)	3.215	0.462	2.308, 4.121	0.000
	9 (164.9 ± 0.4)	−5.130	0.426	−5.965, −4.295	0.000
	10 (188.5 ± 0.5)	−7.180	0.419	−8.002, −6.359	0.000
	11 (212.8 ± 0.3)	−6.274	0.462	−7.179, −5.369	0.000
	12 (235.4 ± 0.6)	−9.230	0.396	−10.005, −8.454	0.000
Intercept		2.997	0.754	1.520, 4.474	0.000

Notes:

^aRegressed onto C_t as the outcome variable.

^bBacteria concentrations were estimated using MPN methods.

^cReferent.

^dBased on the averaged HOBO data collected in 15-min. intervals beginning 24 h prior to harvest.

^eCategorized time since inoculation with actual hours displayed in parenthesis.

^fEvent 7 was modeled as two separate events to capture the 4.5 h collection lag between both trials.