Table 1.

Mean coverage probability of each of the four pointwise 95% confidence intervals (averaged across s), under each scenario. VI and V2 are Wald-type confidence intervals based on the model-based estimates of the variance V₁ and V₂, defined in Section 2.4. B-V is a Wald-type confidence interval based on the variance of the bootstrap estimates, and B-Q is constructed from the 2.5% and 97.5% quantiles of the boostrap distribution.

N	Dataset	Opt. Method	β1(s)				β2(s)				β3(s)
			V1	V2	B-V	B-Q	V1	V2	B-V	B-Q	V1	V2	B-V	B-Q
100	Full	AIC	90	83	100	99	88	82	100	99	76	64	100	99
		AICC	87	81	100	98	86	82	100	98	70	58	99	95
		EPIC	68	63	96	94	82	80	100	96	56	46	93	86
	Inomplete	AIC	63	54	98	98	88	81	100	98	66	41	85	82
		AICC	60	52	98	98	87	80	100	98	64	40	85	81
		EPIC	50	44	96	96	84	79	100	97	55	36	82	77
200	Full	AIC	95	89	100	98	86	81	100	99	80	68	100	97
		AICC	95	88	100	98	86	81	100	98	76	64	99	95
		EPIC	89	80	97	94	79	76	100	96	61	50	86	82
	Inomplete	AIC	78	65	95	98	86	80	99	98	64	37	73	71
		AICC	76	64	95	98	85	79	100	98	63	36	72	70
		EPIC	65	56	91	96	78	75	100	97	53	33	69	67
500	Full	AIC	97	91	100	98	88	83	100	99	83	72	99	96
		AICC	97	92	100	98	88	82	100	99	82	71	99	95
		EPIC	95	87	98	94	79	75	99	96	63	51	80	81
	Inomplete	AIC	92	68	93	97	87	79	99	98	66	27	62	63
		AICC	92	68	93	97	87	79	99	98	66	27	61	63
		EPIC	88	65	90	96	80	74	98	96	56	25	59	60