On the Stata's nonparametric bootstrap technique

[Nader Mehri]

Hi,
I have a question about Stata's nonparametric bootstrap technique (the bootstrap command). Given that the bootstrap samples are supposed to be drawn with replacement, I wonder if we can conclude that the "with replacement" term is redundant in Stata's manual on the nonparametric bootstrap technique. Thanks for your thoughts.

NM

[Carlo Lazzaro]

Nader:
I would keep it in the Stata .pdf manual.
In my opinion, the reason to add "with replacement" to bootstrap replication is useful to distinguish this technique from other simulation approaches, such as Monte Carlo, that run without replacement.
This difference is particularly useful for those who come across these statistic tools for the very first time, whereas, as you say, becomes implicit when the researcher gets more and more familiar with them.
I'd be interested in Felix Bittmann 's opinion about this issue.

[Felix Bittmann]

I suppose the wording in the manual is just for clarification, maybe for people who are not familiar with the method. As the definition of the bootstrap is to be always with replacement ("Bootstrapping is any test or metric that uses random sampling with replacement...", Wikipedia), one could say that the wording is redundant in the manual.