Donald Burrill wrote:
>
> I believe that at one time it may have been believed (or perhaps
more
> correctly, hoped) that knowledge of the first four central moments
> would provide a useful empirical way of describing the shape of an
> empirical distribution, much as a polynomial of degree 4 can
> sometimes provide not too bad a description of many empirical
> functions (over a limited range
> of values, of course).  But in practice this seems never to have
> happened:  perhaps because in practice one is concerned largely with
> the shape of the tails of distributions, not with the shape of their
> central 75% (say).

I think that recent workers would disagree with this in a subtle way.
There has been some work an alternatives to moments as descriptors,
notably L-moments. Papers on L-moments suggest that a major reason for
not using ordinary moments is related to sampling variability: you
need an extremely large sample to get a reasonable estimate of
skewness and kurtosis. Again, there are problems with the skewness
(and kurtosois) cofficient, in that, for a given sample size, the
range of possible sample values is bounded.

A slightly different point is that the "method of moments" is quite
often used in fitting parametric distributions, although often with
poor results.

David Jones


.
.
=================================================================
Instructions for joining and leaving this list, remarks about the
problem of INAPPROPRIATE MESSAGES, and archives are available at:
.                  http://jse.stat.ncsu.edu/                    .
=================================================================

Reply via email to