Hello!
I want to study the power of the Wilcoxon test for the test :
H0 : Fx = Fy
against H1 : Fx = F1 > Fy = F2
F is the empirical distribution function
F1 = F(normal(0,1)) and F2 = F(normal(0.5,1))
here I'm trying to stimulate 50 samples of Z in order to have their p-values.
The aim is to evaluate using Monte Carlo the power of the Wilcoxon test.
Tn <- function (z,n1) {
return(abs(mean(z[1:n1])-mean(z[n1+1:length(z)])))
}
pval <- function (z,B,nX){
boot_sampl = c()
for (i in 1:B){
boot_sampl[i] = Tn(sample(z),nX) # bootstrap samples
}
p = boot_sampl[boot_sampl>Tn(z,nX)]
p_val = mean(p)
return((p_val))
}
Is this correct?
How can I implement the Monte-Carlo method in order to evaluate the power of the Wilcoxon test?
Thank you!