0

I have a computer simulation from which I get a sample of values of some microscopic quantity $X$, i.e. $\{ x_1,\ldots,x_N \}$. I'm interested in estimating the expected value of $X$, i.e. $E[ X]\approx\frac{1}{N}\sum x_i$, and the variance associated with this estimate.

Now, do I understand correctly that:

  1. taking the estimate of the variance of $X$, $Var[X]\approx\frac{1}{N}\sum (x_i-\bar{x})^2$, is NOT what I'm looking for (which in this notation would be $Var[E[X]]$)?

  2. I need to make use of some statistical technique, such as bootstrapping?

whuber
  • 281,159
  • 54
  • 637
  • 1,101
ponadto
  • 171
  • 2
  • 8
  • I am voting to migrate to math.SE! –  Aug 20 '14 at 20:38
  • @Danu probably better for stats SE... can't auto-migrate there, but if ponadto requests it (which I would suggest, since there doesn't appear to be any physics in this question), one of the mods can move it over. – Kyle Aug 20 '14 at 20:41
  • Oh, stats is even better! I forgot it existed... –  Aug 20 '14 at 20:43
  • Sorry, but how can I "request a migration" (gracefully)? – ponadto Aug 20 '14 at 20:45
  • 1
    The duplicate showed up in a site search for ["standard error" + mean](http://stats.stackexchange.com/search?q=%22standard+error%22+mean). Many more related posts can be found in the same way. Please note that "$E[X]$" usually refers to a *number*--the expectation of $X$--and that your formula is the *sample mean*, which is a realization of a *random variable.* It is often written $\bar{X}$ and its realization (as given by the formula) as $\bar{x}$. This question seeks information about computing the standard error of the sample mean *qua* estimator of the expectation. – whuber Aug 20 '14 at 21:31

0 Answers0