Table 8: Type I error and power under DGP2.
|
Method1 |
n0 |
n1 |
p |
θ1 = 1 |
θ1 = 2 |
θ1 = 3 |
θ1 = 4 |
θ1 = 5 |
|
PCGC |
500 |
500 |
10 |
0.05 |
0.056 |
0.052 |
0.05 |
0.06 |
|
PCGC |
1000 |
1000 |
10 |
0.046 |
0.046 |
0.046 |
0.046 |
0.042 |
|
PCGC |
1500 |
1500 |
10 |
0.056 |
0.046 |
0.044 |
0.048 |
0.05 |
|
Fork |
500 |
500 |
10 |
1 |
1 |
0.05 |
0.762 |
1 |
|
Fork |
1000 |
1000 |
10 |
1 |
1 |
0.074 |
1 |
1 |
|
Fork |
1500 |
1500 |
10 |
1 |
1 |
0.064 |
1 |
1 |
|
TW |
500 |
500 |
10 |
0.074 |
0.008 |
0.046 |
1 |
1 |
|
TW |
1000 |
1000 |
10 |
0.192 |
0.026 |
0.062 |
1 |
1 |
|
TW |
1500 |
1500 |
10 |
0.33 |
0.036 |
0.052 |
1 |
1 |
θ1 = 3 constitutes the null distribution
1PCGC: Principal Components Group Comparison; Fork: Forkman’s Test; TW: Tracy-Widom