Computational Statistics

132 COMPARISONS

Example 3.10:

Input

nsimul <- 500

n1<- 10; n2 <-10

alpha <- 0.01

x<-0

for (i in 1:nsimul) {

if (t.test(exp(rnorm(n1)),exp(rnorm(n2, mean = 1)),

alternative="less",

var.equal = TRUE)$p.value < alpha){

x <- x+1}

}

p <- x/nsimul

cat("estim p", p)

Output

estim p 0.188

Input

prop.test(n = nsimul, x = x)

Output

1-sample proportions test with continuity correction

data: x out of nsimul, null probability 0.5

X-squared = 193.442, df = 1, p-value < 2.2e-16

alternative hypothesis: true p is not equal to 0.5

95 percent confidence interval:

0.1552568 0.2256388

sample estimates:

p

0.188

In library(binom) [8] several tools are provided for a differentiated analysis of