The American Statistician for May carried out a note
Jones, M.C. 2004. Hazelton, M.L. (2003), "A graphical tool for assessing
normality," The American Statistician 57: 285-288: Comment. The
American Statistician 58: 176-177.
which prompted me to blow the dust off a token Stata program
I wrote in 2002 to do density probability plots. The original
reference is
Jones, M.C. and F. Daly. 1995. Density probability plots.
Communications in Statistics, Simulation and Computation 24: 911-927.
Thanks to Kit Baum, a greatly reworked version is now
available as -dpplot- from SSC. The original Stata 7 version
is still available as -dpplot7- from the same package
but otherwise frozen as was.
Bail out now unless you want to know a little more
about density probability plots.
The basic idea is to plot two things (1) a smooth
density function obtained in the standard way
by plugging parameter values into a formula
(2) an estimate of the density obtained as the
density as a function of the quantiles as a function
of the (cumulative) probability f(Q(P)), again
given parameter values. In both cases we are
considering the badness of fit of some (continuous)
distribution family (Gaussian, etc.).
The graphs take some getting used to but can
be helpful and attractive. The original -dpplot-
from 2002 supported just fitting normal and
exponential distributions, perhaps not very
interesting examples, as in each case there
are familiar graphical methods readily
available for assessing distribution fit. The
latest version supports more than that and
inspection of the code should make it fairly
clear to user-programmers how to add code
for your own favourite distributions.
There is much more background in the help file.
Nick
n.j.cox@durham.ac.uk
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/