At 07:41 PM 3/19/02 +0000, Joe wrote: >I'm taking an undergrad business stat. course. The prof. gave us a formula >to calculate the coefficient of skewness: > >3(mean - median)/(std. dev )
well, look at this ... in symmetrical distributions ... mean = median so, you would have in your numerator 0 and hence, 0 skewness a very simple formula that works on the same principle is mean - median ... quick and rough, but maybe good enough IF you have to find a skewness value >And told us that the formula that excel uses to calculate skewness is >different/wrong. > >I did a little digging to try and understand why Excel would use a >non-standard formula. The only equation that I found which came close to >what my prof. offered up was Pearson's second coef. of skewness which is: > >3[mean] - [median] / (std. dev.) but, look at this one? i read this as multiplying the mean by 3 first ... then subtracting the ratio of the median divided by the sd ... let's say we had a norm. dist. with mean = 100 and sd = 10 ... we know the median is also 100 so ... 3(100) - (100/10) = 300 - 10 = 290 ... ????????? i don't think your latter formula is anywhere near correct one good way to measure skewness is using a mean of the cubed z scores ... 1. find the zs .. 2. cube the zs 3. take and average i think this is called gamma 1 >Two questions, is the formula my prof. gave me valid, and why wouldn't I >use the Excel calculation for skewness? Is the formula I was given more >applicable to business statistics than what Excel uses? > >Thanks for helping the statistcally insignificant. >-Joe > > >. >. >================================================================= >Instructions for joining and leaving this list, remarks about the >problem of INAPPROPRIATE MESSAGES, and archives are available at: >. http://jse.stat.ncsu.edu/ . >================================================================= Dennis Roberts, 208 Cedar Bldg., University Park PA 16802 <Emailto: [EMAIL PROTECTED]> WWW: http://roberts.ed.psu.edu/users/droberts/drober~1.htm AC 8148632401 . . ================================================================= Instructions for joining and leaving this list, remarks about the problem of INAPPROPRIATE MESSAGES, and archives are available at: . http://jse.stat.ncsu.edu/ . =================================================================
