Table 7.

Out-of-Sample Predictive Accuracy of the Models: Average Values with 95% Confidence Intervals in Parentheses

Bias	MNL	MNLH	MNP	MNPH	SMNP	SMNPH
6 sites	0.3585	0.3227	0.4913	0.4540	0.3550	0.3564
	(0.3663, 0.3503)	(0.3309, 0.3143)	(0.5044, 0.4774)	(0.4678, 0.4395)	(0.3638, 0.3463)	(0.3656, 0.3484)
12 sites	0.3636	0.3854	0.6016	0.5791	0.3830	0.3742
	(0.3750, 0.3521)	(0.3981, 0.3723)	(0.6108, 0.5937)	(0.5883, 0.5682)	(0.3941, 0.3709)	(0.3885, 0.3620)
24 sites	0.4960	0.4156	0.6646	0.6058	0.6098	0.5877
	(0.5042, 0.4863)	(0.4254, 0.4055)	(0.6706, 0.6582)	(0.6152, 0.5997)	(0.6173, 0.6032)	(0.5974, 0.5793)
Mean Squared Error
6 sites	0.1975	0.1672	0.2863	0.2618	0.2008	0.2042
	(0.2017, 0.1930)	(0.1708, 0.1635)	(0.2982, 0.2742)	(0.2715, 0.2509)	(0.2045, 0.1976)	(0.2089, 0.2006)
12 sites	0.2093	0.2294	0.3905	0.3673	0.2287	0.2147
	(0.2162, 0.2028)	(0.2390, 0.2202)	(0.4000, 0.3825)	(0.3764, 0.3581)	(0.2351, 0.2224)	(0.2222, 0.2072)
24 sites	0.3116	0.2571	0.4694	0.3966	0.4221	0.3855
	(0.3173, 0.3046)	(0.2640, 0.2504)	(0.4764, 0.4615)	(0.4077, 0.3914)	(0.4285, 0.4163)	(0.3937, 0.3783)
Predictive Accuracy
6 sites	0.6830	0.8462	0.6650	0.6563	0.7496	0.7461
	(0.7395, 0.6569)	(0.8515, 0.8389)	(0.7155, 0.6485)	(0.6590, 0.6496)	(0.7521, 0.7479)	(0.7510, 0.7270)
12 sites	0.8603	0.7780	0.8566	0.8558	0.8579	0.8610
	(0.8619, 0.8546)	(0.8023, 0.7479)	(0.8609, 0.8441)	(0.8598, 0.8567)	(0.8598, 0.8567)	(0.8630, 0.8598)
24 sites	0.7448	0.8069	0.7263	0.7985	0.7041	0.7877
	(0.7479, 0.7427)	(0.8117, 0.7950)	(0.7395, 0.7176)	(0.8180, 0.7657)	(0.7207, 0.6705)	(0.8096, 0.7657)

Note: The model results with the lowest bias, mean squared error, and predictive accuracy are highlighted in bold. The naïve expectations for the bias are 0.833, 0.917, 0.958 for the 6-site, 12-site, and 24-site models, respectively. The naïve expectations for the mean squared error are 0.694, 0.840, and 0.918 for the 6-site, 12- site, and 24-site models, respectively. Model names are defined in Table 2.