Stata The Stata listserver
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

Re: st: Axis rules made to be broken


From   Constantine Daskalakis <[email protected]>
To   [email protected]
Subject   Re: st: Axis rules made to be broken
Date   Thu, 15 May 2003 10:08:55 -0400

At 06:49 AM 5/15/03, R. Allan Reese wrote:
Apart from the joy of writing that subject line, I feel able to comment
because I have looked at the literature, come to a wider conclusion, and
for some years have taught a module in data graphics.
You actually clarified quite a few things for me. I agree with lots of what you say with the following provisos:

1. Whether one plots on the original or the log or whatever scale is NOT a whim but is dictated by how the audience understands the message. Plotting percentages on a log scale is kind of bizarre -- there is no intuitive scientific reason for doing it (as opposed to plotting, say, some assay values on a log scale, which people normally analyze on the log scale). Almost everyone would be puzzled by it.

2. Maximizing the amount of info one conveys within a fixed-size communication (here, area of graph) is NOT a whim but good practice. Would you write a 100-page book where 9 out of 10 pages are left blank and only one takes up the story you want to tell (and with tiny font, at that)? Why not write the exact same story using all 100 pages, but with a font 10 times as big? Wouldn't that be a more efficient way of communicating the message?

3. I was indeed talking about line graphs (points connected by a live over time), scatterplots, etc, NOT bar graphs.


An axis should contain zero *if the statement being made is one of
absolute magnitude.*

A graph that focuses on *change* need not include the origin of the
absolute figures, because change implies "change from what?" and it is
this reference point that becomes the de facto centre of attention. If
bars are drawn, they should be from the reference point, but a line-plot
may be more effective in emphasizing the direction of change.
I guess you justified my argument better than I did here. :)
In my example, the interest is comparison over time, i.e., change from 1990, as opposed to the actual absolute magnitudes of the percentages. I want to see the fluctuations over time. In other words, I want to take a magnifier, focus it on the region around 70-90%, and see how the trend goes up or down.


CD gives an example of plotting percentages which is particularly
informative. In many situations the interest lies not on x but on 100-x.
"Use of PCs has gone up from 70 to 80%" is equivalent to "one third of
those who did NOT use PCs now do so".
Actually, this is one of many interpretations. My interest is as stated above.





The documents accompanying this transmission may contain confidential health or business information. This information is intended for the use of the individual or entity named above. If you have received this information in error, please notify the sender immediately and arrange for the return or destruction of these documents.
________________________________________________________________

Constantine Daskalakis, ScD
Assistant Professor,
Biostatistics Section, Thomas Jefferson University,
125 S. 9th St. #402, Philadelphia, PA 19107
Tel: 215-955-5695
Fax: 215-503-3804
Email: [email protected]
Webpage: http://www.kcc.tju.edu/Science/SharedFacilities/Biostatistics

*
* 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/




© Copyright 1996–2024 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index