[R] boxplot notches

[Martin Maechler]

>>>>> "P" == P B Pynsent <p.b.pynsent at bham.ac.uk>
>>>>>     on Tue, 2 Mar 2004 15:16:23 +0000 writes:

    P> A Google search showed  that all this was discussed in April 1998 with 
    P> an extensive reply to the question from M Maechler.

Yes, indeed:
     http://finzi.psych.upenn.edu/R/Rhelpold/archive/0839.html , 
and I hadn't remembered,  there giving quite a bit more numeric details
than we have had in this thread. 

    P> I, as a non-statistician, blindly believed what was written in the 
    P> boxplot() help file, I am sure many would be grateful to this help 
    P> being modified.

there's nothing wrong in there, AFAIK, is there?

    P> I still do not understand why , 6 years later with GHz processors, 
    P> boxplot() could not have an option to produce exact intervals. After 
    P> all,  a range option is offered for the whiskers.
    P> At least then non-overlapping notches would have some meaning, wouldn't 
    P> they?

back in 1998, I had answered to Peter Dalgaard's

       PD> Search me... However, wouldn't it be better in any case to do an
       PD> exact 95% CI based on the binomial distribution? Of course, you
       PD> need at least 6 observations to do that.

  MM> No, please not yet another definition of the boxplot!
  MM> People looking at boxplots should be able to rely on their knowledge of
  MM> what a boxplot is.

and I still very much adhere to that.

If one really wants, there's not too much wrong with adding
something like "median.test()" with the corresponding confidence
interval {if it's agreed that you'd want the "close-to-boundary"
	  order statistics and [pq]binomial for that},
but I'd already vote tentatively against another
boxplot option which would change the way the notches are
computed.

Martin Maechler <maechler at stat.math.ethz.ch>	http://stat.ethz.ch/~maechler/
Seminar fuer Statistik, ETH-Zentrum  LEO C16	Leonhardstr. 27
ETH (Federal Inst. Technology)	8092 Zurich	SWITZERLAND
phone: x-41-1-632-3408		fax: ...-1228			<><