topic Re: Design effect interpretation in Statistical Procedures

Design effect interpretation

kmel56 — Thu, 06 Aug 2015 15:18:14 GMT

Hello,

I'am using the Surveyfreq procedure to get design effect of my proportions, but i have a problem with deisgn effects interpretation, see the result below :

DEFF

ALL 1.32

Women 0.7

Men 0.7

my sampling design is a one stage clustering design, so i understand having the 1.32 (>1) but why the deff is so smaller for men an women?

Can you help me please?

Re: Design effect interpretation

Rick_SAS — Fri, 07 Aug 2015 12:53:31 GMT

Are you using the RATE= or TOTAL= option? It would be useful to see your syntax. The way that SURVEYFREQ estimates the design effect is documented as follows:

SAS/STAT(R) 9.3 User's Guide

Re: Design effect interpretation

kmel56 — Thu, 20 Aug 2015 07:46:48 GMT

Thanks Rick,

But the problem was away, I didn't include the weighting factor, now with weights it much better.

Re: Design effect interpretation

Rick_SAS — Thu, 20 Aug 2015 12:38:15 GMT

The lesson here is "always include your program when you ask a question."

Re: Design effect interpretation

kmel56 — Thu, 20 Aug 2015 13:13:33 GMT

Yes, Sorry, my question was incomplete.

THanks.