I need to get empirical size and power of test. Also, I need some R code help. is this right? b = qnorm(1-(0.05/2)))b[i]=1 } sum(b)/nsim

I need to get empirical size and power of test. Also, I need some R code help.

is this right?

b<-numeric(nsim)

for(i in 1:nsim) {

if((chi[i]-E1[i])/sqrt(V1u[i]) >= qnorm(1-(0.05/2)))b[i]=1

}

sum(b)/nsim

Also, how to get empirical covariance estimator for gee in R i.e. cov(yi) = (yi - pi^)(yi - pi^)' then calculate

cov(Y) = diag( cov(y1).... cov(yN) )