RE: AW: st: xttrans accuracy

["Nick Cox" <n.j.cox@durham.ac.uk>]

There is no problem here, except minutely that anyone installing both my -xttrans2- and Maarten's might well get confused. Maarten did what any experienced Stata user would do: realise that tweaking the code for -xttrans- would provide more handles. 

Two larger issues remain, however, that are more interesting: 

1. The official command -xttrans- remains limited by comparison with what its users often expect from it. 

2. Whether rounded fractions sum to 1 exactly not only involves machine precision issues but also provides some mathematical challenges. See for example 

Persi Diaconis and David Freedman 
1979 
On Rounding Percentages
Journal of the American Statistical Association 74: 359-364.

-- and its references. 

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

Maarten buis

--- On Wed, 23/6/10, Martin Weiss wrote:
> Note that there might be copyright issues re NJC`s -ssc d
> xttrans2-, though...

Knowing Nick I am not so worried about a copyright conflict
with him. But from a practical point of view it is not good 
to have multiple programs with the same name floating around
and Nick was first, so I will happily propose to change
my program to -xttrans3-. From an even more practical point
of view, my program is superfluous. Nick's -xttrans2- allows
the -matcell()- option which means that it can leave behind
the matrix of interest.

*
*   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/