=========================================================================
Date:         Fri, 7 Jul 2006 13:23:16 -0500
Reply-To:     "Reutter, Alex" <areutter@spss.com>
Sender:       "SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>
From:         "Reutter, Alex" <areutter@spss.com>
Subject:      Re: PPS sampling
Content-Type: text/plain; charset="US-ASCII"

Using the CSPLAN and CSSELECT procedures in the Complex Samples option is a fairly painless way to handle this in SPSS.  If I correctly understand what you want to do, you will need a "measure of size" variable.  For example:

Case/Pupil      School  SchoolSize
1               1               116
2               1               116
...
116             1               116
117             2               163
118             2               163
...

Then work through the Sampling Wizard.  You should end up with something like:

CSPLAN SAMPLE
  /PLAN FILE='mysample.csplan'
  /METHOD TYPE=PPS_SYSTEMATIC
  /MOS VARIABLE= SchoolSize
  /RATE VALUE=0.25.
CSSELECT
  /PLAN FILE='mysample.csplan'.

The sampling plan is defined by CSPLAN (systematic sampling of 1/4 pupils with probability proportional to school size) and the specifications are saved to the external file mysample.csplan.  CSSELECT uses the information in this file to carry out the actual sampling.  Analysis procedures in the Complex Samples option use the information this file to ensure correct computation of statistics according to the complex sampling plan.

Cheers,
Alex



> -----Original Message-----
> From: SPSSX(r) Discussion [mailto:SPSSX-L@LISTSERV.UGA.EDU] On Behalf Of
> Cathal McCrory
> Sent: Friday, July 07, 2006 10:06 AM
> To: SPSSX-L@LISTSERV.UGA.EDU
> Subject: PPS sampling
>
> I have the enrolment numbers for 3200 schools and I want to select 13,000
> pupils from within these schools on a PPS basis. For example:
>
>                         Pupil                   Cumulative
> School                  Enrolment               Population
> 1                       116                     116
> 2                       163                     279
> 3                       232                     511
> 4                       204                     715
> 5                       274                     989
> 6                       188                     1177
> 7                       210                     1387
> 8                       407                     1794
> 9                       298                     2092
>
> I want to generate a random start within SPSS and select every 4th pupil
> from within the cumulative population total (rather than a set number of
> cases) and have the program iterate until I have drawn my sample of
> 13,000.
> I was wondering whether anyone could give me any pointers (to existing
> routines) or offer any guidance on this. Many thanks.
>