[R] Bootstrapped Tobit regression - get standard error 0...

[Achim Zeileis]

On Mon, 2 Apr 2012, Ravi Kulkarni wrote:

> I am trying to work out a bootstrapped Tobit regression model. I get the
> coefficients all right, but they all have standard error zero. And I am
> unable to figure out why. I know the coefficients are correct because that's
> what I get when do a Tobit (without bootstrapping). Here's my code:
>
> # Bootstrap 95% CI for Tobit regression coefficients?
> library(boot)
> library(AER) # for the Affairs dataset
> data(Affairs)
>
> # function to obtain regression weights
> bs <- function(formula, data, indices) {
>  d <- data[indices,]
>  fit <- tobit(formula, data=Affairs)
>  return(coef(fit))
> }
> # bootstrapping with 1000 replications
> results <- boot(data=Affairs, statistic=bs,
>   R=1000, formula=affairs~age+yearsmarried+religiousness+occupation+rating)
>
> # view results
> results
>
> That gives me the right coefficients, but the SEs are all zero, so I 
> cannot get 95% confidence intervals...

In the AER book in Chapter 7.2 we have an example for a linear regression 
where boot() is used on an lm() model. This can easily be adapted here:

## refit function and bootstrap
refit <- function(data, i) coef(tobit(
   affairs ~ age + yearsmarried + religiousness + occupation + rating,
   data = data[i,]))
m1 <- boot(Affairs, refit, R = 1000)

## original model
m2 <- tobit(affairs ~ age + yearsmarried + religiousness + occupation +
   rating, data = Affairs)

## compare results
m1
coeftest(m2)

Hope that helps,
Z

> Any help is welcome...
>
>  Ravi
>
>
>
>
>
> --
> View this message in context: http://r.789695.n4.nabble.com/Bootstrapped-Tobit-regression-get-standard-error-0-tp4525760p4525760.html
> Sent from the R help mailing list archive at Nabble.com.
>
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
>