I thoroughly agree with Dennis on these points. (And cannot resist
pointing out that none of his remarks indicate any particularly
compelling utility to having and/or using formal coefficients of
skewness or kurtosis. ;-) -- Don.
On Mon, 17 Mar 2003, dennis roberts wrote:
> it "may" (emphasis on may) be helpful on occasion ... to compare two
> distributions in terms of skewness and the eye might not be so sure ...
> perhaps a quantitative value would be of "some" value
>
> however, at least for introductory work ... if the eye can't see it ...
> then the difference is probably not important
>
< snip, a comment of mine >
>
> i think that it can be helpful to know that IF a distribution is
> highly skewed (large n and rather smooth) ... that the mean and
> median won't be the same ... take in a large n distribution of
> salaries ... the mean salary will usually be further UP the scale
> than will be the median ... this could be important IF one were
> trying to make the case that either salaries were good ... or bad
> ...
>
> which measure of average is reported could help or hurt one's case
-----------------------------------------------------------------------
Donald F. Burrill [EMAIL PROTECTED]
56 Sebbins Pond Drive, Bedford, NH 03110 (603) 626-0816
.
.
=================================================================
Instructions for joining and leaving this list, remarks about the
problem of INAPPROPRIATE MESSAGES, and archives are available at:
. http://jse.stat.ncsu.edu/ .
=================================================================
