# T-test on standard deviation samples?

I want to evaluate the means of two samples of standard deviations.

Can I run a simple T-test on the samples of standard deviations? Does it make sense?

• Do you mean samples of standard deviations (as in you only have multiple values of standard deviations for different samples but not the samples themselves) or standard deviations of samples? – David Ernst Dec 9 '16 at 8:21
• Do you have any roughly plausible distributional assumptions for the original values (the ones the standard deviations were calculated on)? – Glen_b -Reinstate Monica Dec 10 '16 at 3:41