TABLE 2.

Scenario 1: Simulated level (Column 4) and power (Columns 5–6) of the new method, that is, the test described in Algorithm 1, considering different sample sizes, censoring rates and thresholds $\mathrm{\Delta }$.

$\left(n_1,n_2\right)$	Censoring rate	Censored (%)	$\mathrm{\Delta }=0.0006$	$\mathrm{\Delta }=0.001$	$\mathrm{\Delta }=0.0015$
(200200)	0.001	25	0.037 (0.000)	0.534 (0.476)	0.964 (0.852)
	0.002	40	0.042 (0.000)	0.410 (0.363)	0.916 (0.752)
	0.003	50	0.053 (0.000)	0.373 (0.313)	0.807 (0.656)
	0.005	63	0.107 (0.000)	0.326 (0.226)	0.772 (0.480)
	0.01	77	0.220 (0.000)	0.290 (0.076)	0.535 (0.209)
(300300)	0.001	25	0.045 (0.001)	0.694 (0.618)	0.990 (0.966)
	0.002	40	0.060 (0.000)	0.587 (0.553)	0.965 (0.932)
	0.003	50	0.056 (0.000)	0.504 (0.470)	0.902 (0.853)
	0.005	63	0.083 (0.001)	0.359 (0.356)	0.772 (0.758)
	0.01	77	0.161 (0.000)	0.307 (0.183)	0.562 (0.452)
(500500)	0.001	25	0.050 (0.001)	0.847 (0.831)	1.000 (1.000)
	0.002	40	0.057 (0.000)	0.791 (0.751)	0.987 (0.985)
	0.003	50	0.052 (0.000)	0.685 (0.682)	0.967 (0.976)
	0.005	63	0.060 (0.000)	0.545 (0.583)	0.887 (0.922)
	0.01	77	0.119 (0.000)	0.371 (0.333)	0.664 (0.772)

Note: The numbers in brackets correspond to the results from the individual procedure. The nominal level is chosen as $\alpha =0.05$. The third column displays the mean proportions of censored individuals.