I like the idea of marking quartiles on a dotplot. I would suggest (a)
using "Q" rather than "Q1" or "Q3" as it is smaller and there
is no chance of confusion; (b) putting these symbols on the axis, so
instead of
> :
> : .
> . . : . : : : :
> . : : : : . : : : : : . . :
> : : : : : : : : : : : : : : : : : :
> : : : : . : : : : : : : : : : : : . : :
> : : : : : : : : : : : : : : : : : : : : :
> : : : : : : : : : : : : : : : : : : : : :
> ---+---------+---------+---------+---------+---------+---C1
> 20.0 24.0 28.0 32.0 36.0 40.0
> Q1 M Q3
putting
:
: .
. . : . : : : :
. : : : : . : : : : : . . :
: : : : : : : : : : : : : : : : : :
: : : : . : : : : : : : : : : : : . : :
: : : : : : : : : : : : : : : : : : : : :
: : : : : : : : : : : : : : : : : : : : :
---+---------+--Q------+-------M-+-------Q+---------+---C1
20.0 24.0 28.0 32.0 36.0 40.0
(note corrected locations! This is *not* just a quibble - I was
able to read off the median and quartiles accurately enough from the
dotplot to notice -without any deliberate attempt to check - that the
values were incorrect [mailer tab interpretation?].
Thus, while your claim
>BUT, they were not designed to report on the median (though sometimes you
>can find it), nor the quartiles (though sometimes you can find them)...
is no doubt technically correct (in terms of "design"), that's not
really the point.)
I must say I don't like the name "doxplot" very much, although I admit
that
Tukey, coiner (it seems) of most of the really flamboyant neologisms in
statistics today, might just have... this is partially because I am a
strong proponent of boxplots for outlier detection, so this device does
not have the "boxplot nature" to my mind.
-Robert Dawson
.
.
=================================================================
Instructions for joining and leaving this list, remarks about the
problem of INAPPROPRIATE MESSAGES, and archives are available at:
. http://jse.stat.ncsu.edu/ .
=================================================================
