[R-sig-eco] bootstrap approach to subsample "pseudo replicates"

[Johannes Radinger]

Hello,

I do have a dataset partly consisting
of possible "pseudo replicates". I'd like
to use a bootstrap approach to assess
the influence of them.

My dataset looks like this minimal example:

y <- c(1,5,6,2,5,10) # response
x1 <- c(2,12,8,1,16,17) # predictor 1
rep <- c(1,2,2,3,4,4) # pseudo replicates

df <- data.frame(y,x1,rep)

model <- lm(y~x1, data=df)

The variable rep represents kind of pseudoreplicates.
I'd like to randomly select for each bootstrap step
only one case per number (1,2,3,4). So I want to
select always one case per group of rep. I hope you understand
what I mean. In the end I'd like to get the %variation in
the regression parameter estimates.

How can I do that using boot together with lm?

I had already a look in help(boot), but it is still very confusing
to me. Hopefully you can help me.



Best regards,
Johannes


--