RE: [S] Pie Diagrams

Peter Bouman (pbouman@sachs.com)
Fri, 13 Nov 1998 09:30:35 -0600


I would say that even non-scientists in the audience (I am not a scientist,
but a commercial data analyst)
have an interest in clear presentation of data.

Working in the marketing industry, I see large amounts of information
presented in graphical form-often in pie charts.
Usually, the graphics have been prepared with something like Excel or
Powerpoint. The pie charts or overcomplicated bar
charts (with too many dimensions crossed together) that result always make
it difficult to interpret relationships between the data points given,
and could usually be given in a simpler form.

So-I wrote my first message not merely for theoretical reasons, but because
I would like to see wider usage of simpler graphical methods (bar charts,
tables, etc.)

Even Dr. Gunter's e-mail points out a simpler way (without the need for
color) than pie charts to display large datasets. Would that more people
in industry thought the same way.

> -----Original Message-----
> From: Gunter, Bert [SMTP:bert_gunter@merck.com]
> Sent: Friday, November 13, 1998 8:40 AM
> To: 'Garrett, Robert'; 's-news'
> Subject: RE: [S] Pie Diagrams
>
> Dr. Garrett:
>
> It is wonderful to hear from a REAL scientist with REAL scientific
> problems.
> Thankyou for your note to s-news. I hope that it encourages other silent
> scientific readers to speak up!
>
> May I suggest an alternative display that might do better than your pies,
> however?
>
> At each location display the following small rectangular graphic:
>
> The left part of the graphic is a narrow fixed width bar whose height
> represents the total mineral weight at the site. Hence comparing these
> heights by eye gives a more accurate and still geographically located view
> of the total weights than comparing circles (whose sizes we judge as
> areas,
> not diameters, so it is important to encode the total as sqrt(diameter)
> not
> diameter, even in the pies).
>
> (Clearly separated with a small break or vertical line), the right part of
> the graphic (perhaps in a different color, though this is not necessary)
> consists of a (filled?) curve: The "x-axis" is simply the first 5
> different
> minerals, always in the same order of course (the 6th is redundant, but
> you
> could include it if you wanted). The height at each mineral type is simply
> the proportion of minerals at the site. The curves give a more accurate
> and
> easier to perceive view of the mineral "profiles" both because it is
> easier
> to make comparisons on a fixed linear scale and because the positions of
> each of the mineral proportions in the graphic do not vary from site to
> site
> as they do for your pie chart; changing positions makes it more difficult
> to
> make comparisons (that is why you have to add color cues -- but color cues
> are always problematic because we perceive different colors differently.
> One
> slice may look bigger simply because it is a different color).
>
> So not only can you do better than pie charts, you also don't need colors
> (not everybody has color copiers yet!). Of course, there may be times when
> pie charts are the best solution, but I don't think this is it, yet.
>
> Again, I appreciate your writing in. I always find honest science
> more interesting than complex statistical ruminations.
>
>
> > Bert Gunter
> > Biometrics Research RY 70-38
> > Merck & Company
> > P.O. Box 2000
> > Rahway, NJ 07065-0900
> > Phone: (732) 594-7765 Fax: (732) 594-1565
> >
> > "The business of the statistician is to catalyze the scientific learning
> > process." -- George E.P. Box
> -----------------------------------------------------------------------
> This message was distributed by s-news@wubios.wustl.edu. To unsubscribe
> send e-mail to s-news-request@wubios.wustl.edu with the BODY of the
> message: unsubscribe s-news
-----------------------------------------------------------------------
This message was distributed by s-news@wubios.wustl.edu. To unsubscribe
send e-mail to s-news-request@wubios.wustl.edu with the BODY of the
message: unsubscribe s-news