Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down at the end of May, and its replacement, statalist.org is already up and running.


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

re: st: quantile-quantile plots


From   "Ariel Linden, DrPH" <ariel.linden@gmail.com>
To   <statalist@hsphsun2.harvard.edu>
Subject   re: st: quantile-quantile plots
Date   Tue, 16 Apr 2013 10:03:13 -0400

Nick did not mention his own program -qplot- (findit qplot), which improves
on the standard -qqplot- that comes with Stata.

As for "Y versus X", I would venture to guess it is more of an American
thing that British. I suppose the logic is that X and Y are correlated but
not necessarily causal or temporally aligned, thus  you can change them
around and get the same result. 

Conversely, when you say "Y is a function of X" it implies directionality
and even causality...

Ariel 

Date: Mon, 15 Apr 2013 11:21:12 +0100
From: Nick Cox <njcoxstata@gmail.com>
Subject: st: quantile-quantile plots

Some recent threads have mentioned quantile-quantile plots. Q-Q plots
go back to the nineteenth century in the specific case of so-called
normal probability plots (really observed quantiles versus expected
normal quantiles, or vice versa,  depending on what one means by
versus(*)).

The key paper mentioned by David Hoaglin

Wilk MB, Gnanadesikan R. 1968. Probability plotting methods for the
analysis of data.  Biometrika 55:1-17.

is over 40 years old, but the idea is still not yet anywhere as widely
practised as it should be. By the way, that is a spectacular paper,
which pays re-reading every few years.

I want to push a few related hints that might otherwise be buried in
those threads, which people interested in this topic might well not be
following.

First off, and most obviously, Stata has long had a -qqplot- command.

However, it assumes that you want to compare two variables. If not,
- -cquantile-  (SSC) is a helper command that gets you to that
situation.

Quantile-quantile plots are for comparing two distributions. If you
have several, then -qplot- (SJ) allows overlay, or display
side-by-side.

If you want yet more ideas in the same vein, check out

SJ-7-2  gr0027  . .  Stata tip 47: Quantile-quantile plots without
programming
        . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  N. J.
Cox
        Q2/07   SJ 7(2):275--279                                 (no
commands)
        tip on producing various quantile-quantile (Q-Q) plots

free .pdf
http://www.stata-journal.com/sjpdf.html?articlenum=gr0027

Nick
njcoxstata@gmail.com

* A note on "versus" or "against". I was brought up -- probably in
school physics, but it was last century, so I can't give precise
recollections -- to say "plot y versus x" where y is, not
surprisingly, whatever is plotted on the y or vertical axis. That
seems to me to match mathematics and physics usages such as y is a
function of x, or output is a function of time, where the dependent
variable (outcome, response) is always mentioned first. But I've come
across people saying "plot x versus y". If anyone has a logic for that
usage, I'd be interested to hear it.
*


*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/


© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   Site index