[Julia Mueller]

Thanks Nick and Asjad for helping me out, I appreciate it!

Indeed, Nick is correct in #15:

triplot belief1 belief2 belief3 [fw=count],  frame(lpat(dot)) ms(O) y(lpat(solid) lwidth(vthin))

produces the graph below and instead of the lines, I'd love either the markers to be of different colors or the background to be colored differently. Thanks!



Edit: Actually, considering the fact that some markers might end up right on the y()-line, different background colors probably makes more sense.

[Nick Cox]

Different marker colours should be easy, with something like

	
	
gen which = 1
replace which = 2 if belief2 > belief1 & belief2 > belief3 
replace which = 3 if belief3 > belief1 & belief3 > belief2

and separate(which) in the triplot call. I imagine it's equally easy with ternary.

Background colours remains more work, but each zone is just a polygon whose vertices need to be specified.

The coding problem falls into a marginal zone for me as not trivial but not interesting me much compared with many other things I need to do. .