Ricardo Ovaldia wrote:
Is there a way to perform an equivalence test for a 2x2 table.
Treatment (0,1) vs. Outcome(0,1)?
--------------------------------------------------------------------------------
You can use either -cs-, which is official, or -rdci-, which is user-written and
available on SSC. They both provide confidence intervals for a difference in
independent proportions. The latter provides several alternative methods for
risk-difference confidence intervals that have been touted as better in some way
to the conventional method used by -cs-. The former also provides confidence
intervals for odds ratio, if that's what you're using as the statistic.
One of the methods employed in -rdci- relies on -ridder-, another user-written
command that must be installed before using -rdci-. -ridder- has just been
updated, so -findit ridder- and make sure to update the copy that might be
already installed on your machine.
You would construct the confidence interval and check to see whether both lower
and upper confidence bounds fall within the limits specified for accepting the
alternative hypothesis of therapeutic equivalence.
Joseph Coveney
version 11.0
clear *
set more off
set seed `=date("2009-08-13", "YMD")'
set obs 200
generate byte treatment = runiform() > 0.5
generate int success = rbinomial(1, 0.7)
cs success treatment
rdci success treatment
exit
*
* 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/
