Robust method for estimating standard deviation confidence intervals of non-normal data
Nov 2, 2016 5:07 AM(1497 views)
I am looking for a method of estimating standard deviation confidence intervals of non-normal data, e.g. the D. G. Bonett method:
"Approximate confidence interval for standard deviation of nonnormal distributions", Douglas G. Bonett, doi:10.1016/j.csda.2004.10.003
but other methods to solve the problem will do just fine also. I can't find any such method in JMP. Am I missing it, or is it relly not there (then consider this a feture request)? Does anyone know of a script or addin to do it?
An alternative question might be, is the "standard" standard deviation the measure the correct measurment of variaiblty for my data? In distribution, red triangle>display options>customize summary statistics there are some alternatives you might want to explore (look towards the bottom of the list).
But back to your question, if you have JMP Pro you can on-click bootstrap to get a bootstrap confidence interval. Alternatively, you can use the formula editor to enter the formula for the CI you mention.