HP Forum Archive 15

[ Return to Index | Top of Index ]

Puzzling HP-49G+ SDEV-function Message #1 Posted by Karl-Ludwig Butte on 26 Mar 2006, 7:03 a.m. Hello All, I have a difficult question about the SDEV function of the HP-49G+ which puzzles me: There are 4 figures for which I would like to calculate the standard deviation: 6.5 12.1 5.4 -10.3 I've read this excercise in a financial mathematic text book and the author is calculating the standard deviation as follows: SDEV = SQRT(1/4*(SQR(6.5-3.4)+SQR(12.1-3.4)+SQR(5.4-3.4)+SQR(-10.3-3.4)) SDEV = 8.3 When entered into the HP-49G+ and using the formula above the same result is displayed. But using the predefined function Mean and SDEV the 49 displays: Mean: 3.4250 SDEV: 9.6088 First I thought it maybe because in the textbook the 3rd and 4th decimal isn't used so I recalculated with 3.4250 instead of 3.4 but in vain. Tries with an HP-49G, HP-48G and HP-41 the same differences in results were displayed. So it doesn't seem to be a bug of the HP-49G+ Firmware (Version: HP49-B Rev.: 2.01-2). Has anyone of you a hint where the difference comes from ? Thank you very much in advance. Has anyone

Re: Puzzling HP-49G+ SDEV-function Message #2 Posted by Arnaud Amiel on 26 Mar 2006, 8:03 a.m., in response to message #1 by Karl-Ludwig Butte If your 4 figures are the whole population then 1/4 should be used in your formula. However, the 49 is assuming you are using a sample of 4 elements to get an estimate of the standard deviation of a total population in which case it correctly uses 1/(n-1) formula. If you want to use the 1/n SDEV for a whole population you have to use the PSDEV command. See p584 of the PDF user guide for an hint or look a the very good advanced user guide which has better desciptions p248 and p279. Arnaud Edited: 26 Mar 2006, 8:46 a.m.

Re: Puzzling HP-49G+ SDEV-function Message #3 Posted by Karl Schneider on 26 Mar 2006, 8:12 p.m., in response to message #1 by Karl-Ludwig Butte Karl-Ludwig -- Arnaud is correct. HP has generally favored providing the function for sample standard deviation (SDEV) instead of population standard deviation (PSDEV). The workaround for models without a PSDEV function is to enter the mean as an additional datum, and recalculate the sample SDEV. -- KS

Re: Puzzling HP-49G+ SDEV-function Message #4 Posted by Karl-Ludwig Butte on 27 Mar 2006, 4:25 a.m., in response to message #1 by Karl-Ludwig Butte Thank you both for your very good answers. I wasn't aware of this distinction. You helped me a lot - thanks again.

[ Return to Index | Top of Index ]

Go back to the main exhibit hall