Inference on the standard deviation for non-normal distributions

Hello all,

I wanted to ask how do I make inference (confidence intervals, hypothesis testing, etc...) on the the standard deviation or variance, when my distribution is non-normal ?

In particular, I have two cases: A right-skewed distribution and a bimodal distribution that seems to come from two normal distributions.

To my understanding, in both cases inference on the mean is easy, since my sample is large (approx. 60-70 cases). But what about the standard deviation ?

Thanks !


Omega Contributor
I don't know, when you look at the means do they seem appropriate given graphical expressions of data? The typical approach is to present medians with interquartile ranges or attempt to normalize these data. Just make sure you are not misrepresenting data parameters.


Ambassador to the humans
If you go Bayesian there isn't really any added complication. Otherwise you could just bootstrap to get a confidence interval.