twipc_means: Simulation-Based Power Analysis for Two-Way Interaction Using...

In markhwhiteii/twipc: Intuitive Simulation-Based Two-Way Interaction Power Calculation

Description

Estimate power for the interaction term in a 2 x 2 design. This function uses mean patterns to do so. It simulates data from normal distributions where the standard deviations are 1. This means that mean differences between cells can be interpreted in the units of Cohen's d. Thus, one can hypothesize a Cohens'd d in one condition, a Cohen's d in another condition, and then run simulations to estimate power at different sample sizes. See examples.

Usage

twipc_means(a1b1, a1b2, a2b1, a2b2, start = 100, end = 500, by = 25,
  alpha = 0.05, reps = 1000, verbose = FALSE)

Arguments

a1b1	Mean in cell A1, B1
a1b2	Mean in cell A1, B2
a2b1	Mean in cell A2, B1
a2b2	Mean in cell A2, B2
start	Smallest sample size N to test
end	Largest sample size N to test
by	The incrementing size from start to end
alpha	The alpha level for determining significance
reps	The number of times to repeat the simulations
verbose	Logical. Print out current status in the simulation?

Value

A data.frame containing columns indicating sample size and power at that sample size.

Examples

# imagine d = .5 in condition A and d = .0 in condition B
# try see what power is at 100, 150, and 200 N
set.seed(1839)
twipc_means(0, .5, 0, 0, 100, 200, 50)

markhwhiteii/twipc documentation built on Nov. 4, 2019, 5:25 p.m.