Bias at two different medically important levels.

Table 5: Comparison in different sample sizes (Skewed-normal).

prev = 0.55, sen = spec = 0.85, w₀ = 0.85, EF = 0.81, α = -1

Setups	Parameters	True	Empirical	NP Bayes	MLE (EM)
n₁ = 30 $μ_{T}$ $α_{T} () \sim N (3.52, 0.89)$ $α_{C} () \sim N (3.02, 0.89)$	$μ_{T}$	3.71	3.7252	3.6858	3.7603
	$μ_{C}$	3.23	3.2153	3.2025	3.2070
	q	0.48	0.5099	0.5098	0.5533
	$σ_{C}^{2}$	0.35	0.3481	0.3421	0.3071
	$σ_{T}^{2}$	0.3754	0.3897	0.3729	0.3564
	$ω_{1}$	0.8738	0.85	0.9003	0.8468
	Reject H₀	-	0.2655	0.2834	0.3273
	${\hat{n}}_{2} (M e d / M e a n / T r i m)$	-	(60,281,95)	(60,174,79)	(60,301,84)
n₁ = 70 $α_{T} () \sim N (5.47, 0.87)$ $α_{C} () \sim N (4.34, 0.87)$	$μ_{T}$	5.21	5.2311	5.2378	5.3036
	$μ_{C}$	4.51	4.5205	4.5154	4.5126
	q	0.70	0.7106	0.7224	0.7910
	$σ_{C}^{2}$	0.74	0.7455	0.7094	0.6890
	$σ_{T}^{2}$	0.7940	0.8177	0.7832	0.7942
	$ω_{1}$	0.8738	0.85	0.8525	0.8130
	Reject H₀	-	0.5409	0.5868	0.6148
	${\hat{n}}_{2} (M e d / M e a n / T r i m)$	-	(140,219,145)	(140,194,144)	(140,210,143)
n₁ = 90 $α_{T} () \sim N (9.93, 1.01)$ $α_{C} () \sim N (9.93, 1.01)$	$μ_{T}$	9.82	9.8425	9.8459	9.9307
	$μ_{C}$	9.04	9.0478	9.0474	9.0390
	q	0.78	0.7947	0.7970	0.8917
	$σ_{C}^{2}$	0.93	0.9247	0.8871	0.8618
	$σ_{T}^{2}$	0.9971	1.0132	0.9763	0.9994
	$ω_{1}$	0.8738	0.85	0.8482	0.8092
	Reject H₀	-	0.6866	0.7086	0.7565
	${\hat{n}}_{2} (M e d / M e a n / T r i m)$	-	(180,221,183)	(180,211,182)	(180,214,184)