Table 1 Simulation results of the FWER for normal distributed data. This table shows simulation results of the relevance-shifted Westfall-Young permutation algorithm using 500,000 permutation runs ('permutation') and the procedure with a data-driven order of relevance-shifted hypotheses ('selector') for different levels of sample sizes, variances and correlation coefficients using normal distributed data.

${\theta }_{lower}^{-1}$ = θ _upper

n _i

σ _ij

ρ

weak

strong

weak

strong

1

5

10

0.1

3.42

3.05

0

0

1.5

5

10

0.1

2.53

1.98

0

0

5

5

10

0.1

2.53

1.87

0

0

1

10

10

0.1

4.39

4.22

4.65

4.12

1.5

10

10

0.1

2.70

2.81

3.30

3.14

5

10

10

0.1

2.70

2.48

3.30

3.01

1

10

10

0.5

4.26

4.22

4.74

4.61

1.5

10

10

0.5

3.08

3.07

3.36

3.71

5

10

10

0.5

3.08

2.89

3.36

3.59

1

5

10

0.9

3.12

3.13

1.42

0.91

1.5

5

10

0.9

2.83

2.61

1.37

1.45

5

5

10

0.9

2.83

2.80

1.37

1.44

1

5

15

0.1

3.40

2.52

0

0

1.5

5

15

0.1

2.48

1.73

0

0

5

5

15

0.1

2.48

1.98

0

0

1

10

15

0.1

4.27

4.21

4.65

4.10

1.5

10

15

0.1

2.67

2.78

3.30

3.11

5

10

15

0.1

2.67

2.51

3.30

2.97

1

20

10

0.9

5.01

4.96

4.90

4.58

1.5

20

10

0.9

4.04

3.97

3.29

3.64

5

20

10

0.9

3.90

3.97

3.29

3.64