fwerPowerFdrPower: Simulate FWER, POWER, FDR, and POWER
In mshasan/OPWpaper1: Reproduce figures and simulations of OPWeight

Description Usage Arguments Details Value Author(s) References See Also Examples

This function simulate the Family Wise Error Rate (FWER) and the corresponding Power, and the False Discovery Rate (FDR) and the corresponding Power for the different effect sizes

1
2
3

fwerPowerFdrPower(i, simu, null, corr = 0, cv = 0, alpha = 0.05,
  groupSize = 100, effectType = c("continuous", "binary"),
  covariateEffectVec, datWeightByNull)

`i`	Integer, i-th effect size of a vector of effects
`simu`	Integer, number of replications
`null`	Numeric, proportion of the true null hypothesis
`corr`	Numeric, correlation between the test statistics
`cv`	Numeric, coefficient of variation of the test statistics
`alpha`	Numeric value of the significance threshold
`groupSize`	Integer, number of test statistics per group
`effectType`	Character ("continuous" or "binary"), type of effect sizes
`covariateEffectVec`	A numeric vector of different covariate-effect size
`datWeightByNull`	A numeric matrix of weights, each column corresponds to an covariate-effect size

This function simulate Family Wise Error Rate (FWER) and corresponding Power, and False Discovery Rate (FDR) and the corresponding Power for the different effect sizes

A matrix of 16 rows containing information about FWER, POWER, FDR, and POWER (4 rows for each item)

Mohamad S. Hasan, shakilmohamad7@gmail.com

Hasan and Schliekelman (2017)

weight_byEffect_cont ranksProb_byEffect

# vector of covariate-effect sizes
covariateEffectVec <- c(1, 1.5, 2)

# compute probability matrix
ranksProb_byEffect <- sapply(1:length(covariateEffectVec), ranksProb_byEffect,
             null = .9, m = 100, covariateEffectVec = covariateEffectVec)

# compute weights
weightByEffect <- sapply(1:length(covariateEffectVec), weight_byEffect_cont,
                   alpha = .05, null = .9, m = 100, delInterval = .01,
                   covariateEffectVec = covariateEffectVec,
                   datByNull = ranksProb_byEffect)

simuVal = 3  # in the actual case use at least simVal = 1000
result <- sapply(1:length(covariateEffectVec), fwerPowerFdrPower, simu = simuVal,
             null = .5, corr = 0, cv = 0, alpha = .05, groupSize = 100,
             effectType = "continuous", covariateEffectVec = covariateEffectVec,
             datWeightByNull = weightByEffect)