Suppose we don't know anything about a distribution except that it is a Normal distribution with mean $ \mu $ and variance $ \sigma^2 $.

Can we argue, theoretically, the sample variance, $ S_n^2 $ will be close to the population variance, and thus can we say something about the ratio $ \frac {S_n^2}{\sigma^2} $?

If this holds no basis in theory -- can we at least argue this in practice (mostly)?


Normal distribution is fully defined by its mean and variance, that is, if you know $\mu$ and $\sigma^2$ you know everything about the distribution.

The data points in the sample are however drawn randomly, so the sample variance will be close to $\sigma^2$ in probabilistic sense. In the case of a normal distribution $S_n^2$ follows a $\chi^2$-distribution with $n-1$ degrees of freedom.

I suggest looking for derivations of Student t-test, which usually cover the relation between real and measured means and variances.

| cite | improve this answer | |

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.