送交者: 秦州刺史 于 2006-4-05, 14:00:41:
回答: 新到中xj关于error bar的解释有一问题 由 littlebird 于 2006-4-05, 11:13:02:
just like Rodger said:
"It is the same to prove sum of squares is smaller than
square of sum, given all elements are larger than 0."
But I didn't find a proof for Mean-SE>=X(min) and from
my simple Excel calculation, I found a counter example:
0.1, 0.1, 0.1, 9.e+13
Mean-SE=0.09765625, which is < 0.1.
Here, SE=SD/SQRT(N), and SD refers to the bias-corrected
variance,just like how Mathematica implements StandardDeviation[list]
Could this be due to numerical uncertainty?