TABLE V

Performance Metrics for Comparison of H and HK1 Versus GPQ and BayesPP

Method	SMALL Sample				LARGE Sample
Method	Mean^{^a} LRL	CC LRL	Mean^{^a} LSL	CC LSL	Mean^{^a} LRL	CC LRL	Mean^{^a} LSL	CC LSL
BayesPP (fixed slope)	0.9891	0.89	0.9877	0.89	0.9974	0.73	0.9969	0.80
BayesPP (random slope)	0.9892	0.86	0.9650	1.00	0.9974	0.70	0.9948	0.89
GPQ	0.9836	0.95	0.9827	0.96	0.9923	0.95	0.9918	0.96
H	0.9799	0.97	0.9781	0.97	0.9901	0.98	0.9891	0.99
HK1	0.9828	0.95	0.9812	0.97	0.9925	0.95	0.9919	0.95

↵a Mean limits are normalized to true values. CC = confidence coefficient; LRL= lower release limit; LSL = lower shelf-life limit.