powcorplot: Plotting of power curves I

In AlexanderRitz/pairedpowR: Power Estimation for paired data

Description

Plotting of estimated power curves of unpaired and paired t-test in relation to underlying group correlation. In case of no actual mean difference, i.e. effect = 0, the curves represent the observed rate of type I errors and should be interpreted as the significance level of the tests.

Usage

powcorplot(
  gran = 100,
  numpairs = 10,
  effect = 0,
  diffdisplay = FALSE,
  mod = FALSE
)

Arguments

gran	integer. Supplies the level of granularity, i.e. the number of points for interpolation of the final curve.
numpairs	integer. Supplies the number of pairs to be simulated. Resulting in two samples of size "npairs".
effect	numeric. Supplies the difference in means between both groups. Can be interpreted as effect size based on the fact that standard normal variates are the basis of simulation.
diffdisplay	logical. If TRUE, plots will show difference between the realised power or significance levels of paired and unpaired test. The difference is calculated by subtracting the values of the unpaired test from the paired tests values.
mod	logical. If TRUE, plots will show the realised power or significance levels of paired and modified paired tests. The difference for diffdisplay = TRUE is calculated by subtracting the values of the paired test from the modified paired tests values.

Value

Returns plots of either the power curves(or significance levels) of paired and unpaired t-test in relation to the underlying group correlations, or alternatively the plotted difference between both curves.

Examples

powcorplot(gran = 10, numpairs = 5, effect = 0.5)

AlexanderRitz/pairedpowR documentation built on March 16, 2020, 12:16 a.m.