Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.
From | "Airey, David C" <david.airey@Vanderbilt.Edu> |
To | "statalist@hsphsun2.harvard.edu" <statalist@hsphsun2.harvard.edu> |
Subject | Re: st: power calculation graphs |
Date | Fri, 20 Aug 2010 09:48:39 -0500 |
. While you can use -postfile-, you can also work directly with a Stata data file. Here's a simple example of making a power graph: clear set obs 7 // make a data set of 7 observations generate size = . // prepare variable for data generate power = . // prepare variable for data local i 1 // i is used to identify data rows foreach power of numlist 0.50(0.10)0.90 0.99 { // set up a loop sampsi 9 13, sd1(2) sd2(4) power(`power') alpha(0.05) replace size = r(N_1) in `i' // add size result to data set replace power = `power'*100 in `i' // add power result to data set local i = `i' + 1 // increment to the next row } // end the loop graph twoway connected power size, /// xlabel(0(5)25) mc(red) lc(red) // make a graph graph export plot1.pdf, replace // Apple OS X only Also, there are power(ful) articles in the Stata Journal: Generalized power calculations for generalized linear models and more R. Newson. 2004. Stata Journal Volume 4 Number 4. Power by simulation A. H. Feiveson. 2002. Stata Journal Volume 2 Number 2. On Fri, Aug 20, 2010 at 8:54 AM, Nikolaos Pandis <npandis@yahoo.com> wrote: > Hi to all. > > I was wondering if someone would be able to help me draw the following sample calculation graphs with Stata: > > 1. Two arm parallel trial, outcome binary. > > A power (y-axis) sample size (x-axis) calculation graph for various failure proportions between trial arms. For example, p1=0.80 and p2=0.85, or p1=0.80 p=0.90 or p1=0.80 p=0.95 > > > 2. Two arm parallel trial, outcome continuous. > > A power (y-axis) sample (x-axis) calculation graph for various failure proportions between trial arms. For example, m1= 100 and m2= 120 with equal sd1=sd2=10. It would be great if I could vary m1, m2, sd (always sd1=sd2) > You can do both of these using the -sampsi- command and could have found this yourself by typing -search sample size- into Stata. If you want to do the analysis for a range of values then you should consider putting each call within a series of nested -forval- loops and write the results to a file using -postfile- which you can then read into Stata and draw the graphs you want. I'll leave the actual coding of this as an exercise for you to work on ;-) Neil * * 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/