0

The technique of bootstrapping, a technique used to describe estimators using sample data, is described by Greene (Econometric Analysis, 8th ed.) as it follows:

enter image description here

In such technique, sample size ($m$), as a percentage of $Z$, may be as small as 0% and as large as 100%. How should we choose it?

Incognito
  • 111
  • 3
  • 1
    The text clearly states the sample size is $m:$ there is no choice. I am hoping the duplicate addresses any remaining questions about the possibilities of other resample sizes. If not, then see https://stats.stackexchange.com/questions/tagged/bootstrap?tab=Votes. – whuber Feb 09 '21 at 17:52
  • I don’t understand what you mean. m = 0 will produce one result, m = 50% will produce another, and m = 100% would produce yet another. For the bootstrapping, we may pick any of them, but they are not equally efficient. Which one is the best? – Incognito Feb 09 '21 at 18:00
  • You misunderstand. A bootstrapped sample *must* obtain $m$ values, as stated. Note that $m$ is a count, not a percentage (and 0% would correspond to a count of zero in any case). Please see the related posts for clarification. – whuber Feb 09 '21 at 18:02
  • OK, I meant m as percentage of Z (closest to possible in natural numbers). – Incognito Feb 09 '21 at 18:04

0 Answers0