|Re: HP 17BII Sigma calculations are inconsistent|
Message #2 Posted by Thibaut.be on 20 May 2001, 3:00 a.m.,
in response to message #1 by Michael Burns
The answer is yes and no.
Remember your maths classes ?
S is the square root of the square of the sum of the differences with the mean divided by the sample (n)
But to get ONE difference, you need TWO values, to get 5 differences, you need 6 values. So statisticians developped (I don't know the exact word in English) a "corrected" standard deviation which is the same formula, but divided by n-1 instead of by n.
The largerpopulation, the less effect n or n-1 has on the standard deviation calculation.
I don't know the 17BII, but I do know the 19BII and no doubt that the used formula is this corrected standard deviation (divided by n-1). You want to get the normal standard deviation ? Nothing more easy : once your list entered, calculated the mean, and key in S+ (sigma plus). You'll add 1 to n, so n-1+1 = n, and the difference of this item with the mean is null... This is even explained in the 19BII manual, so I guess this problem and how to treat is is also explained in the 17BII manual.
By the way, I am just creating a website about my HP collection, but also with the limits of use HP financial calcs and how to bypass them. You are all encouraged to visit hpcalc.multimania.com and leave me feedback to my site still in construction.