2
$\begingroup$

In my literature book I found that the z-score can be calculated by the following function:

z = (data point - mean) / standard deviation

Now I'm reading on bootstrap and I found that they calculated the z-score using the outcome of the boostrap analysis by the following R code:

$tm$ is a vector holding the difference in response times between two groups.

$diff.mean = function(x,i)$ $mean(x[i])$

$boot(tm,diff.mean,R=20000)$ $$ z = tm.boot$t0/sd(tm.boot$t) $$ Can someone explain why the mean is not used in this formula? Is it because the mean is 0? If so, why is the mean 0?

Improve this question
$\endgroup$
5
  • $\begingroup$ If you don’t provide more context, it will be difficult to answer... what is bootstrapped? $\endgroup$
    – Elvis
    Mar 14, 2015 at 15:38
  • $\begingroup$ I run the bootstrap equivalent to a paired t-test by using a vector which holds the difference of two vectors. I'm bootstrapping to find a difference between two groups in terms of the mean. $\endgroup$
    – Daemonstool
    Mar 14, 2015 at 15:53
  • $\begingroup$ Doesn’t the paired $t$-test bit explain why $0$ is taken as the expected value? In fact your bootstrap is used only to estimate a standard deviation. $\endgroup$
    – Elvis
    Mar 14, 2015 at 16:07
  • $\begingroup$ So the mean is 0 as the expected difference between the two groups are 0? $\endgroup$
    – Daemonstool
    Mar 14, 2015 at 16:12
  • 1
    $\begingroup$ yes it looks like it's because it's testing against 0 (it's like they're doing [score-0]/SE, where SE is the standard error) $\endgroup$
    – Patrick Coulombe
    Mar 14, 2015 at 16:26

0

Reset to default

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.