Bootstrap regression model, Resample the cases.


library(bootstrap)
data(cell)
plot(cell) 

ffit <- lm(log.surv ~ dose + I(dose^2), data=cell)
summary(ffit)
curve(predict(ffit, data.frame(dose = x)), add = TRUE)

alpha.hat <- coef(ffit)[1]
beta1.hat <- coef(ffit)[2]
beta2.hat <- coef(ffit)[3]

x <- cell$dose
y <- cell$log.surv
n <- nrow(cell)
nboot <- 1000
alpha.star <- double(nboot)
beta1.star <- double(nboot)
beta2.star <- double(nboot)

for (i in 1:nboot) {
    k.star <- sample(x=1:n, size=n, replace = TRUE)
    x.star <- x[k.star]
    y.star <- y[k.star]
    bfit <- lm(y.star ~  x.star + I(x.star^2))
    alpha.star[i] <- coef(bfit)[1]
    beta1.star[i] <- coef(bfit)[2]
    beta2.star[i] <- coef(bfit)[3]
}

beta1.hat
sd(beta1.star)
beta2.hat
sd(beta2.star)
beta2.hat / sd(beta2.star)


Try the same code but with lm replaced by rlm( )
and use psi=psi.huber
(load library(MASS) first )