Skip to main content

Table 5 The empirical type I error rates of the tests at the significance level 0.05 when the survival times have uniform, exponential, log-normal,uniform and exponential, uniform and log-normal, or exponential and log-normal distributions with equal medians but unequal variances and sample sizes n 1 = 20, n 2 = 25, n 3 = 25, n 4 = 30

From: Comparing survival curves based on medians

Distributions and parameters Censoring rate BC TJ New
Uniform distribution
θ = (10,10,10,10)
a = (4,4,4,4)
c = (2,3,4,5)
0 0.087 0.062 0.067
0.1 0.087 0.075 0.057
0.2 0.087 0.086 0.051
0.3 0.099 0.118 0.053
Exponential distribution
θ = (10-2log(2),10-3log(2),10-4log(2),10-5log(2))
a = (1/2,1/3,1/4,1/5)
0 0.084 0.062 0.064
0.1 0.086 0.067 0.060
0.2 0.089 0.078 0.062
0.3 0.100 0.100 0.054
Log-normal distribution
θ = (log(10), log(10), log(10), log(10))
a = (2,3,4,5)/10
0 0.078 0.057 0.060
0.1 0.088 0.077 0.064
0.2 0.090 0.093 0.063
0.3 0.087 0.109 0.052
Two uniform and two exponential distributions
θ = (10, 10, 10-10*log(2),10-10*log(2))
a = (0.1,0.1,0.1,0.1)
c = (0.01,0.01,0.01,0.01)
0 0.074 0.052 0.058
0.1 0.088 0.089 0.049
0.2 0.093 0.106 0.046
0.3 0.092 0.119 0.039
Two uniform and two log-normal distributions
θ = (10,10, log(10), log(10))
a = (5,5,0.3,0.3)
c = (2,2,2,2)
0 0.076 0.056 0.058
0.1 0.085 0.083 0.050
0.2 0.088 0.105 0.049
0.3 0.097 0.132 0.047
Two log-normal and two exponential distributions
θ = (log(10), log(10),10-log(2)/0.3,10-log(2)/0.3)
a = (0.3,0.3,0.3,0.3)
0 0.068 0.049 0.051
0.1 0.078 0.062 0.055
0.2 0.083 0.080 0.054
0.3 0.096 0.111 0.059