Testing for differences in distribution tails to test for differences in 'maximum' lifespan

Table 3 Performance of the tests in simulation 3, H _0,Bis true, H _0,Ais false and f (Y|Y ≤ τ ∩ X = 1) is radically different from f (Y|Y ≤ τ ∩ X = 0) (see Figure 3 for details of simulation).

Test	Sample Size (N) Per Group
	50		80		100
	α = .05	α = .01	α = .05	α = .01	α = .05	α = .01
*Tests of H* _{0, A} *(Wang-Allison tests)*
QT3 with τ set to 130.	0.244	0.101	0.412	0.181	0.490	0.258
QT3 with τ set to sample 90^th percentile.	0.102	0.102	0.332	0.051	0.297	0.143
QT4 with τ set to 130.	0.266	0.102	0.418	0.187	0.514	0.274
QT4 with τ set to sample 90^th percentile.	0.102	0.102	0.332	0.151	0.297	0.143
*Tests of H* _{0, B}
Wilcoxon-Mann-Whitney with τ set to 130.	0.046	0.013	0.049	0.011	0.045	0.008
Wilcoxon-Mann-Whitney with τ set to sample 90^th percentile.	0.048	0.019	0.044	0.01	0.041	0.009
Permutation test with τ set to 130.	0.042	0.007	0.046	0.012	0.064	0.013
Permutation test with τ set to sample 90^th percentile.	0.046	0.009	0.046	0.012	0.044	0.01
*Tests of H* _{0, C} *(new tests)*
Wilcoxon-Mann-Whitney with τ set to 130.	0.276	0.111	0.420	0.201	0.517	0.271
Wilcoxon-Mann-Whitney with τ set to sample 90^th percentile.	0.182	0.07	0.278	0.104	0.35	0.154
Permutation test with τ set to 130.	0.291	0.101	0.427	0.203	0.515	0.28
Permutation test with τ set to sample 90^th percentile.	0.169	0.067	0.264	0.107	0.363	0.173

ISSN: 1471-2288