Table 2 Type I error rates and power for treatments effects. The type I error rates and power were evaluated under five different error distributions – normal, uniform, lognormal, contaminated normal and Cauchy. The values inside and outside parenthesis are type I error rates and power, respectively. The EG(PFT) and EG(MRT) denote the efficiency gain of PFT and MRT versus FT.

N

5

0.732 (0.050)

0.732 (0.050)

0.691 (0.054)

0.000

-0.056

10

0.976 (0.050)

0.976 (0.050)

0.966 (0.050)

0.000

-0.010

U

5

0.575 (0.051)

0.565 (0.047)

0.486 (0.048)

-0.017

-0.155

10

0.930 (0.054)

0.930 (0.054)

0.876 (0.050)

0.000

-0.058

LN

5

0.386 (0.029)

0.461 (0.047)

0.587 (0.048)

0.194

0.521

10

0.576 (0.036)

0.620 (0.052)

0.889 (0.052)

0.076

0.543

CN

5

0.565 (0.039)

0.598 (0.058)

0.736 (0.055)

0.058

0.303

10

0.829 (0.047)

0.838 (0.056)

0.958 (0.052)

0.010

0.156

C

5

0.230 (0.021)

0.366 (0.054)

0.564 (0.049)

0.591

1.452

10

0.259 (0.016)

0.402 (0.052)

0.848 (0.055)

0.552

2.274