Extending gridlines in triplot

[Tom Snell]

Hi,

I am using triplot from SSC in STATA15.

The triangular plot is ideal for my purposes, but difficult to read since there is an ambiguity as to which gridline corresponds to each axis.

Is there an option to extend the gridlines (maintaining their angle) beyond each axis? Example attached, with the suggested lines in red.

If not, this would be a very useful (and seemingly vital) addition.

Many thanks

[Nick Cox]

Thanks for your interest, but there is no such option and there are no plans to provide one. Sorry, but I have never felt the need for this or even seen it in print.

Although I am quite fond of this command, these plots have limitations too. One is that many people are unfamiliar with them, although not Earth scientists and some kinds of economists and political scientists, for example.

Another limitation is that they don't always make good use of the space. (Special plots showing just part of the space are common in Earth science.)

A canonical example is US election data where Republicans, Democrats and all others are the usual choices. For such data, "all others" is typically small and the other variables often close to 50% so only some of the space is used effectively.

Democrats MINUS Republicans or its negation

versus

others

is a much easier plot for most people to think about. It's a distortion of the simplex, which some might care about. I don't know what your data are but

(A MINUS B) versus C

is what I would try. given that C is often ~10%. When A MINUS B is in effect the response of interest, the point is intensified.

I gave a talk on these ideas and other ideas in 2008 in London. The slides are at https://www.stata.com/meeting/uk08/abstracts.html but I put together a more accessible version in 2010 which I am happy to send to anyone intensely interested if they email me. It is not especially hard to find an email address for me but I am coy about putting any of mine in yet another place. If there is more interest than say 3 people I will post the document on SSC.

[Tom Snell]

Thanks Nick - the London slides are very useful for exploring alternatives.

Tom