Statalist


[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

st: -spineplot- update available from SSC


From   "Nick Cox" <n.j.cox@durham.ac.uk>
To   <statalist@hsphsun2.harvard.edu>
Subject   st: -spineplot- update available from SSC
Date   Tue, 6 Nov 2007 14:38:30 -0000

Thanks to Kit Baum, the -spineplot- package on SSC has been updated. 
-spineplot- implements spine plots for two-way categorical data. 
Stata 8.2 is required. Use -ssc- to install or update if interested. 

In this version there are various changes, as in part previously 
promised: 

1. A -text()- option allows showing text at the centre of each tile. 
This grows out of a suggestion on Statalist by Peter Jepsen, to whom 
thanks for a good idea (which I initially resisted). 

2. -aweights- are now supported. 

3. Several detailed changes have been made to the help, which is 
quite detailed (20 references). 

The immediate trigger for writing this program, as has been previously 
reported, was a Statalist posting by Matthias Schonlau. Further 
off-list discussions with him and with Antony Unwin have underlined, 
to my surprise, that the original definition of spine plots was much 
more restrictive than is implemented here.  This was in a paper by 
Hummel (1996), which I cite in the help: so far no paper or electronic
version has come to my hand. However, offering a program that can draw a

wider variety of graphs than the original definition seems no great 
vice to me, and I don't feel obliged to follow the original narrow 
definition slavishly when a wider definition is very useful. However, 
on some terminology what -spineplot- draws are not just spine plots 
strict sense, but rather mosaic plots for two variables. 

I did also look at the implementation of spine plots in R, which 
at a wild guess is the software most frequently used for drawing them, 
and it takes a similar stance. 

I've kept the name -spineplot-, not least because the only obvious 
alternative -mosaicplot- would be even more misleading, as the program 
does not implement mosaic plots as usually understood for three or 
more variables. I've no ambitions to extend the program that far. 
I've yet to see a mosaic plot for so many variables that didn't 
seem to cry out for a simpler alternative, ingenious though the idea
remains. 

Nick 
n.j.cox@durham.ac.uk 


*
*   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 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index