Sampling with different sampling rates from each strata

[Mike Smet]

Hello,
I have a dataset from which I have to take a stratified sample in order to conduct a survey.
I want to take different samples from each strata: e.g. 10% from strata A, 20% from strata B and 40% from strata C.
I tried both the 'sample' command, as well as 'gsample', however it seems that using these commands the sampling rate has to be equal for all strata. Is this correct? How can I specify different sampling rates for each strata?

Thanks,
Kind regards,
Mike

[Mike Lacy]

Here's an example of -gsample- doing what you want:

 
// example data
clear
set obs 10000
gen int stratum = ceil(runiform() * 3)    
gen int pct = stratum * 10  // for example
//
gsample pct, strata(stratum) gen(insample) wor percent
tab insample stratum, col

For a small demonstration of technique, here's the do-it-yourself approach:

gen pct .....
gen rand = runiform()
sort stratum rand
by stratum: gen byte insample = (_n <=(_N * pct/100))

[Clyde Schechter]

From the helpfile for -gsample- (which, by the way is not official Stata and is available from SSC):

gsample [#|varname] [if] [in] [weight] [, options]
...
For stratified sampling, # units will be selected from each stratum identified by the strata() option. Alternatively, specify varname instead of #, where
varname is a variable containing for each stratum a specific sample size. varname is assumed to be constant within strata.

Added: Crossed with #2 who points out the same thing, in greater detail.

[Mike Smet]

Thanks a lot for your quick responses: I simply didn't realise that I could use a variable name instead of a number after the gsample command. Sorry for the inconvenience ...

[Feruzbek Davletov]

Thank you very much for all responses!