Table 2.

The case when we observe p = 0.001. Sample output from the R script, calc-FPR + LR.R (see electronic supplementary material). Values mentioned in the text in red.

INPUTS

true mean for sample 1 = 0

true mean for sample 2 = 1

true s.d. (same for both samples) = 1

Observed p-value = 0.001

Calculation of FPR for specified n

OUTPUTS for false pos risk, with: prior, p(H1) = 0.1

CASE OF p = alpha

For nsamp = 4 False positive risk = 0.526 power = 0.0089

Lik(H1/Lik(H0) = 8.12

For nsamp = 8 False positive risk = 0.208 power = 0.0497

Lik(H1/Lik(H0) = 34.3

For nsamp = 16 False positive risk = 0.0829 power = 0.238

Lik(H1/Lik(H0) = 99.6

OUTPUTS for false pos risk, with: prior, p(H1) = 0.5

CASE OF p = alpha

For nsamp = 4 False positive risk = 0.110 power = 0.0089

Lik(H1/Lik(H0) = 8.12

For nsamp = 8 False positive risk = 0.0289 power = 0.0497

Lik(H1/Lik(H0) = 34.3

For nsamp = 16 False positive risk = 0.00994 power = 0.238

Lik(H1/Lik(H0) = 99.6