Re: st: RE: Venn diagrams

[David Elliott <dcelliott@gmail.com>]

I have used Jens Lauritson's -venndiag- program in the past and have
been pleased at what it can do.  I have also worked on a spreadsheet
program to calculate properly scaled circles for a three-way
intersection within the Venn universe and was amazed at the amount of
trigonometry necessary to calculate the areas of overlap and the
lengths of the common chords, and to do so with three pairs of
simultaneous equations.

Personally, I find Venn diagrams to be highly intuitive and useful for
illustrating overlapping proportions conceptually to a statistically
unsophisticated audience.  However, they don't have the intrinsic
worth of showing you something that can't be tabulated or graphed in
any other manner.

As an aside, this article [
http://www.ecmaj.ca/cgi/reprint/156/10/1387 ] includes a properly
proportioned diagram I made using my spreadsheet approach after
creating the proportions table based on -venndiag- output.  At the
time this was printed, it was rare for that journal to use colour
graphics but they recognized how well it clarified the situation of
multiple drug use as discussed in the paper.

Also, for those who don't know, Jens has headed up a team at EpiData
Software [ http://www.epidata.dk/ ] that has created a data entry
front-end that provides sophisticated validation, lookups etc. and
import/export to Stata.

DC Elliott

On 22 October 2010 10:31, Nick Cox <n.j.cox@durham.ac.uk> wrote:
> Not so; -venndiag- from SSC is more recent and it supersedes anything in the STB.
>
> But why be surprised?
>
> 1. Writing a Venn diagram program that always looks good and works well is a tricky problem. Wanting the areas exactly right is an easy desideratum to state, but it does not make the problem easier.
>
> 2. Jens Lauritsen hasn't found time to port this to the new Stata graphics; like everybody else he is a very busy person.
>
> 3. Not many users have the Stata programming skills to write something as good or better, and it seems that none of them have the inclination to do this.
>
> 4. StataCorp can be as capricious as any bunch of academics in not doing stuff that doesn't interest them. They can speak for themselves, or not, but I don't sense that this particularly appeals to anyone at StataCorp.
>
> 5. Some may share my own personal view that while Venn diagrams are interesting mathematically and in the history of ideas they are of rather little use in data analysis.
>
> All likely to be very disappointing to anyone frustrated, no doubt, but surprising? I suggest not.
>
> (Michael can also speak for himself, but I think that his book does not purport to survey user-written programs at all.)
>
> Nick
> n.j.cox@durham.ac.uk
>
> Philip Burgess
>
> I'm trying to graph a Venn diagram using (STB-54)  gr34.3
>
> When I run - venndiag var1 var2 - it produces the tabular output
> correctly but crashes altogether when I gues it is trying to graph the
> output.
>
> I'm surprised that - findit venn - indicates that gr34.3 (2000) is the
> 'most' recent ado for Venn diagrams. I could find no reference to such
> in Michael Mitchell's (excellent) VGSG.
>
> I wonder whether there is an appropriate ado ora work around. Ideally,
> I want to generate areas that are poprtioned to the actual data.
>
> I'm using Stata/SE 11.1 (born 30 Sept 2010)
>
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/statalist/faq
> *   http://www.ats.ucla.edu/stat/stata/
>



-- 
David Elliott

Everything is theoretically impossible, until it is done.
Progress is made by lazy men looking for easier ways to do things.
 -- Robert A. Heinlein (American science-fiction Writer, 1907-1988)

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/