st: Re: Simplification of mata syntax to determine CI possible?

[Dirk Enzmann <dirk.enzmann@uni-hamburg.de>]

In reply to Nick ( 
http://www.hsph.harvard.edu/cgi-bin/lwgate/STATALIST/archives/statalist.1107/date/article-184.html 
):

Yes,

pp = range(0.000001,0.999999,1/99999)

is a dodgy thing the do but this it not essential - I did it to compare 
the results experimentally to another routine where I used a vector of 
100,000 rows. Of course, you could work with 100,001 rows and use

 pp = range(0,1,.00001)

, instead.

And yes, I read the Brown et al. (2001) paper. As to the likelihood 
ratio interval they write (p. 114): "Brown, Cai and DasGupta (1999) show 
by analytical calculations that this interval has nice properties. 
However, it is slightly harder to compute." (unfortunately their 
reference to Brown et al., 1999, is lacking).

Because -ci- lacks this type of interval, I tried it on my own. 
Ultimately, I would like to be able to replicate the examples given by 
Brown et al. (2001) using Stata instead of the R (see:

http://www2.jura.uni-hamburg.de/instkrim/kriminologie/Mitarbeiter/Enzmann/Software/prop.CI.r

and

http://www2.jura.uni-hamburg.de/instkrim/kriminologie/Mitarbeiter/Enzmann/Software/ex_prop.CI.r

). However, as I said, I would like to know whether it is possible to 
replace the 13 lines of mata by something more efficiently as the one 
line I used in the R function. Or if there is an even better approach to 
construct the likelihood ratio confidence interval using Stata.

References:
Brown, L.D., Cai, T.T., & DasGupta, A. (2001). Interval estimation for a 
binomial proportion. Statistical Science, 16(2), 101-133.

========================================
Dirk Enzmann
email: dirk.enzmann@uni-hamburg.de
========================================
*
*   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/