TABLE 4.

Simulation results regarding the empirical type I error (H ₀) and the empirical power (H ₁) of the proposed two‐sample Kolmogorov‐Smirnov–type tests for $H_0:P_{0,12}(·)=P_{0,22}(·)$ and $H_0:P_{0,12}^{\prime }(·)=P_{0,22}^{\prime }(·)$ at the $\alpha =0.05$ level

		$P_{0,p2}(·)$, $p=1,2$				$P_{0,p2}^{\prime }(·)$, $p=1,2$
		H 0		H 1		H 0		H 1
n	$F_M$	IF	CB	IF	CB	IF	CB	IF	CB
20	$\mathcal{U}[5,15]$	0.045	0.042	0.352	0.339	0.049	0.050	0.331	0.337
	$\mathcal{U}[10,30]$	0.040	0.039	0.634	0.625	0.044	0.040	0.598	0.601
40	$\mathcal{U}[5,15]$	0.044	0.041	0.666	0.659	0.037	0.039	0.612	0.603
	$\mathcal{U}[10,30]$	0.048	0.046	0.905	0.906	0.044	0.046	0.874	0.873
80	$\mathcal{U}[5,15]$	0.048	0.046	0.916	0.917	0.049	0.047	0.870	0.864
	$\mathcal{U}[10,30]$	0.053	0.053	0.995	0.994	0.059	0.055	0.991	0.990

Abbreviations: $F_M$, distribution of the cluster size; n: number of clusters.

Note. Significance levels were calculated based on either the estimated processes ${\widehat{C}}_{n,2}$ and ${\widehat{C}}_{n,2}^{\prime }$ (IF) or the nonparametric cluster bootstrap (CB). Results under right censoring.