Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.
| From | Nick Cox <n.j.cox@durham.ac.uk> |
| To | "'statalist@hsphsun2.harvard.edu'" <statalist@hsphsun2.harvard.edu> |
| Subject | st: RE: Re: sampclus formula |
| Date | Tue, 17 Jan 2012 19:05:14 +0000 |
On -sampsi-: this is an official command and the underlying methods and formulas are detailed at [R] sampsi.
On -sampclus-: this is a user-written command.
STB-60 sxd4 . . . . . . Sample size estimation for cluster designed samples
(help sampclus if installed) . . . . . . . . . . . . . J. M. Garrett
3/01 pp.41--45; STB Reprints Vol 10, pp.387--393
estimates sample sizes required for the difference of means
or proportions, adjusted for cluster size and intraclass
correlation
The complete text of the article is at
http://www.stata.com/products/stb/journals/stb60.pdf
In case of doubt the code is short and neat and thus also accessible as a record of procedure.
Nick
n.j.cox@durham.ac.uk
cristian larroulet
Here at J-PAL Latin America we are working on some power calculations
that mix the presence of baseline measurements and clustered design.
We are having problems fitting calcs done by excel with what Stata
says (and Optimal design as well). This only happens when we use the
options of clustering (usign sampclus) and "correlation between
baseline and follow-up". The basic cases (only clustering or only
using baseline) work well. It seem to us that Stata is doing an
aditional correction when both options are active. Does anyone here
knows the exact formula that stata has for sampsi-sampclus?
*
* 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/