Table 1.

True and false positive rates (TPR, FPR) for the six scenarios of toxicity/efficacy curves in Figure 2 under 18 combinations of dose-assignment mechanism, efficacy analysis, and 15-patient DECs, 30-patient DECs with no futility analysis, or 30-patient DECs with a futility analysis

	TPR						FPR
	CRM	Global	Local	CRM	Global	Local	CRM	Global	Local	CRM	Global	Local
# in DEC=15
	Set 1			Set 2			Set 1			Set 2
Model	98	87	81	87	72	69	53	53	50	53	54	53
Empiric	90	86	74	77	72	61	41	52	46	43	53	49
	Set 3			Set 4			Set 3			Set 4
Model	97	95	91	93	88	80	74	80	71	80	81	73
Empiric	94	96	93	90	89	84	76	83	79	77	83	79
	Set 5			Set 6			Set 5			Set 6
Model	93	79	73	86	72	61	51	52	48	48	45	37
Empiric	85	78	65	78	71	51	39	51	44	39	44	34
# in DEC=30 (no futility analysis at 15)
	Set 1			Set 2			Set 1			Set 2
Model	96	76	63	78	46	34	14	17	14	20	19	15
Empiric	94	76	64	77	46	35	18	19	18	18	19	18
	Set 3			Set 4			Set 3			Set 4
Model	81	75	69	75	45	31	21	19	16	25	20	17
Empiric	78	79	75	73	48	36	21	24	23	21	24	23
	Set 5			Set 6			Set 5			Set 6
Model	89	62	47	76	58	35	15	17	13	12	15	9
Empiric	87	62	49	76	58	35	17	18	17	17	17	12
# in DEC=30 (with futility analysis at 15)
	Set 1			Set 2			Set 1			Set 2
Model	93	75	58	72	44	31	13	16	12	16	17	13
Empiric	85	74	53	64	44	26	12	16	13	13	17	13
	Set 3			Set 4			Set 3			Set 4
Model	83	75	64	73	45	30	20	18	15	25	19	16
Empiric	76	79	68	69	49	32	19	23	20	20	23	20
	Set 5			Set 6			Set 5			Set 6
Model	87	62	43	75	59	37	13	16	12	11	14	9
Empiric	79	62	39	70	58	31	12	16	13	12	14	9

TPR (left-hand columns) is the probability that a dose was recommended in an efficacious DEC, i.e. a DEC having at least one dose level with sufficiently large response rate. FPR (right-hand columns) is the probability that a dose was recommended in an inefficacious DEC, i.e. a DEC having no dose levels with sufficiently large response rate.