2
$\begingroup$

The empirical rule for a normal distribution suggests that 68% of data will fall within the first standard deviation, 95% within the first two standard deviations, and 99.7% will fall within the first three standard deviations of the mean. If I have a data set where all the elements fall within first two standard deviations of the mean, can I consider this data set as a normally distributed one?

$\endgroup$
  • $\begingroup$ What is the size of the sample? $\endgroup$ – Sasha Sep 12 '12 at 4:13
  • $\begingroup$ Isn't there a general rule for this scenario? I have a sample of 100 instances in the current problem that I am dealing with. $\endgroup$ – LVJung Sep 12 '12 at 4:22
1
$\begingroup$

The naive approach would be to expect 5 samples of 100 outside 2 SD, with a standard deviation on that of $\sqrt 5 \approx 2.2$, so getting 0 is only a 2 SD event (and maybe not that bad as the measured SD is probably a bit low).

$\endgroup$
  • $\begingroup$ Are you calculating with $\sigma$ or with $s_{n-1}$? I guess an unusual sample with all data within $\pm2s_{n-1}$ rather indicates that $s_{n-1}$ is unusually high ... ? $\endgroup$ – Hagen von Eitzen Sep 12 '12 at 4:56
0
$\begingroup$

A normally distributed random variable has a pdf with support over all of $ \mathbb R$. If it does not it is some sort of truncated distribution. Check the properties of a truncated normal distribution on Wikipedia to verify if your data satisfies those.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.