Table 9: Type I error and power under DGP3.
|
Method1 |
n0 |
n1 |
p |
φ1 = −0.4 |
φ1 = −0.2 |
φ1 = 0 |
φ1 =0 .2 |
φ1 = 0.4 |
|
PCGC |
500 |
500 |
10 |
0.824 |
0.282 |
0.04 |
0.28 |
0.82 |
|
PCGC |
1000 |
1000 |
10 |
0.936 |
0.382 |
0.05 |
0.416 |
0.944 |
|
PCGC |
1500 |
1500 |
10 |
0.986 |
0.466 |
0.034 |
0.508 |
0.992 |
|
Fork |
500 |
500 |
10 |
1 |
0.984 |
0.06 |
0.984 |
1 |
|
Fork |
1000 |
1000 |
10 |
1 |
1 |
0.052 |
1 |
1 |
|
Fork |
1500 |
1500 |
10 |
1 |
1 |
0.052 |
1 |
1 |
|
TW |
500 |
500 |
10 |
1 |
0.916 |
0.03 |
0.892 |
1 |
|
TW |
1000 |
1000 |
10 |
1 |
1 |
0.042 |
1 |
1 |
|
TW |
1500 |
1500 |
10 |
1 |
1 |
0.026 |
1 |
1 |
φ1 = 0 constitutes the null distribution
1PCGC: Principal Components Group Comparison; Fork: Forkman’s Test; TW: Tracy-Widom