SimTtestPower: Two-Sample Power Calculation Based on Simulation

In FeronG/ClassExample:

Description

Power is the probability we reject the null hypothesis given it is false. Building on this definition, create a function that uses simulations to estimate power for a two-sample T-test.

Usage

SimTtestPower(Var1mean = NULL, Var2mean = NULL, Var1sd = NULL,
  Var2sd = NULL, Var1samplesize = NULL, Var2samplesize = NULL,
  nsim = 100, alphalevel = 0.05)

Arguments

Var1mean	Variable 1 mean
Var2mean	Variable 2 mean
Var1sd	Variable 1 standard deviation
Var2sd	Variable 2 standard deviation
Var1samplesize	Variable 1 sample size
Var2samplesize	Variable 2 sample size
nsim	Number of simulations
alphalevel	alpha-level (default=0.05)

Details

First, you will need to simulate two normally distributed variables, each with a distinct sample size, mean, and standard deviation, and perform a T-test. For that single simulation, evaluate if we would reject the null hypothesis given a specific alpha-level. Now repeat this simulation many times. Power can then be estimated as the proportion of simulations for which we rejected the null hypothesis.

Value

Empirical power calculation

Examples

SimTtestPower(Var1mean=20,Var2mean=22,Var1sd=4,Var2sd=6,
Var1samplesize=40,Var2samplesize=40,nsim=10000,alphalevel=0.05)

FeronG/ClassExample documentation built on March 19, 2018, 4:13 a.m.