Hu, Zidek and Kalbfleisch linear regression bootstrap

library(bootstrap)
data(cell)

ffit=lm(log.surv~dose+I(dose^2),data=cell)
ffit

  #Call:
  #lm(formula = log.surv ~ dose + I(dose^2), data = cell)
  #
  #Coefficients:
  #(Intercept)         dose    I(dose^2)  
  #   1.26911     -1.40242      0.05416 

names(ffit)
  #[1] "coefficients"  "residuals"     "effects"       "rank"         
  #[5] "fitted.values" "assign"        "qr"            "df.residual"  
  #[9] "xlevels"       "call"          "terms"         "model"   

myresid=ffit$resid
sum(myresid)
  #[1] -4.024558e-16       ### should be zero

xmat=as.matrix(ffit$model)
xmat[,1]=1
xmat
  #   log.surv   dose  I(dose^2)
  #1         1  1.175   1.380625
  #2         1  1.175   1.380625
  #3         1  2.350   5.522500
  #4         1  2.350   5.522500
  #5         1  4.700  22.090000
  #6         1  4.700  22.090000
  #7         1  4.700  22.090000
  #8         1  7.050  49.702500
  #9         1  7.050  49.702500
  #10        1  9.400  88.360000
  #11        1  9.400  88.360000
  #12        1  9.400  88.360000
  #13        1 14.100 198.810000
  #14        1 14.100 198.810000

xtxinv=solve(t(xmat)%*%xmat)
zmat=xmat*myresid

n=1000
output=matrix(0,ncol=3,nrow=n)

for (i in 1:n) {
k.star=sample(1:14,size=14,replace=TRUE)
szstar=colSums(zmat[k.star,])
output[i,]=xtxinv%*%szstar
}

output[1:10,]
  #             [,1]        [,2]         [,3]
  #[1,] -0.3271456  0.28759741 -0.023883144
  #[2,] -0.7978400  0.36638753 -0.025507206
  #[3,] -0.3781951  0.21880334 -0.020112902
  #[4,]  0.0589859  0.11459679 -0.020461785
  #[5,] -0.6772981  0.23360651 -0.016703131
  #[6,]  1.4241394 -0.71322810  0.067530879
  #[7,] -0.1868855  0.01149351  0.002674990
  #[8,]  0.4019711 -0.04870342 -0.006039779
  #[9,]  0.7026226 -0.27483582  0.008215798
  #[10,] -0.7730005  0.03410025  0.011512503

covmat=var(output)
c(covmat[1,1],covmat[2,2],covmat[3,3])^0.5
  #[1] 0.5511719 0.2778584 0.0237509

summary(ffit)
  #
  #lm(formula = log.surv ~ dose + I(dose^2), data = cell)
  #
  #Residuals:
  #     Min       1Q   Median       3Q      Max 
  #-1.98443 -0.46592  0.03765  0.54980  2.77457 
  #
  #Coefficients:
  #            Estimate Std. Error t value Pr(>|t|)   
  #(Intercept)  1.26911    1.05542   1.202  0.25443   
  #dose        -1.40242    0.33260  -4.217  0.00144 **
  #I(dose^2)    0.05416    0.02144   2.526  0.02816 * 
  #---
  #Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1 
  #
  #Residual standard error: 1.353 on 11 degrees of freedom
  #Multiple R-Squared: 0.8271,     Adjusted R-squared: 0.7957 
  #F-statistic: 26.32 on 2 and 11 DF,  p-value: 6.415e-05

We see that the estimated standard errors are similar, particularly for beta2, the coef for I(dose^2). 
They are less similar for intercept.