[volt-nuts] 7061 statistics , dont understand

[ed breya]

The discrepancy may be due to the small sample size. I believe you 
want to use the "student's-T" distribution, which has more dispersion 
to account for less information. If you do the experiment with more 
samples - say at least thirty - the distribution should approach 
normal. This information was stored somewhere in my brain and just 
popped into my head when I looked at the post - I haven't looked at 
this sort of thing in many years, so my recollection may not be exactly right.

Ed