Skip to main content

Table 1 Comparison of false positive rate and statistical power for fixed α. For a given significance level (α) the observed false positive rate (FP) and power were determined by simulation using either a point estimate for the variance (PE) or the marginal posterior distribution for Δμ (MP). Values are given for different sample sizes (n1 = n2 = n) and prior degrees of freedom (ν0).

From: A two-sample Bayesian t-test for microarray data

n1 = n2 = n

ν0

αPE

αMP

FPPE

FPMP

PowerPE

PowerMP

2

0

0.05

0.05

-

0.0509

-

0.2169

2

2

0.05

0.05

0.0044

0.0490

0.0450

0.3343

2

4

0.05

0.05

0.0146

0.0535

0.1814

0.3919

2

6

0.05

0.05

0.0211

0.0518

0.2747

0.4182

2

16

0.05

0.05

0.0385

0.0498

0.4186

0.4709

3

0

0.05

0.05

0.0030

0.0500

0.0709

0.4664

3

2

0.05

0.05

0.0167

0.0486

0.2833

0.5404

3

4

0.05

0.05

0.0261

0.0556

0.4153

0.5834

3

6

0.05

0.05

0.0280

0.0500

0.4838

0.6053

3

16

0.05

0.05

0.0389

0.0501

0.5951

0.6441