Alternative Confidence Interval Methods Used in the Diagnostic Accuracy Studies

Table 3

The results of all confidence interval methods for different sample sizes.

	Nickel Sulfate (Ni); ()		Potassium Dichromate (Cr); ()		Lanolin Alcohol (La); ()
	se1 = 0.508	se2 = 0.672	se1 = 0.357	se2 = 0.607	se1 = 0.000	se2 = 0.125
	θ	CI	θ	CI	θ	CI
Asymptotic intervals
Wald	0.164	(0.087, 0.241)	0.25	(0.090, 0.410)	0.125	(−0.104, 0.354)
Wald.cc	0.164	(0.079, 0.249)	0.25	(0.054, 0.446)	0.125	(−0.229, 0.479)
Agresti	0.161	(0.084, 0.238)	0.23	(0.068, 0.398)	0.100	(−0.170, 0.370)
Bonett-Price	0.164	(0.083, 0.240)	0.25	(0.056, 0.411)	0.125	(−0.234, 0.434)
Conditional intervals
Exact.cond	0.164	(0.085, 0.203)	0.25	(0.045, 0.250)	0.125	(−0.119, 0.125)
Exact.mid	0.164	(0.093, 0.200)	0.25	(0.076, 0.250)	0.125	(−0.113, 0.125)
Unconditional interval
Uncond	0.164	(0.090, 0.245)	0.25	(0.111, 0.428)	0.125	(−0.142, 0.445)
Score intervals
Wilson	0.164	(0.086, 0.238)	0.25	(0.086, 0.388)	0.125	(−0.215, 0.471)
Wilson.cc	0.164	(0.083, 0.241)	0.25	(0.070, 0.400)	0.125	(−0.294, 0.533)
Wilson.phi	0.164	(0.085, 0.239)	0.25	(0.071, 0.400)	0.125	(−0.215, 0.471)
Tango	0.164	(0.090, 0.247)	0.25	(0.099, 0.434)	0.125	(−0.240, 0.471)
Nonparametric methods based on relative effects
np.nv	0.164	(0.087, 0.241)	0.25	(0.087, 0.413)	0.125	(−0.120, 0.370)
np.	0.164	(0.087, 0.241)	0.25	(0.083, 0.47)	0.125	(−0.171, 0.421)