Date:         Tue, 18 Apr 2006 16:31:42 -0700
Reply-To:     David L Cassell <davidlcassell@MSN.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         David L Cassell <davidlcassell@MSN.COM>
Subject:      Re: QR: proc surveylogistic
In-Reply-To:  <200604181044.k3I7Cf7C022682@mailgw.cc.uga.edu>
Content-Type: text/plain; format=flowed

adel_tangi@YAHOO.FR wrote:
>I would like to ask what will be the difference between
>
>cluster, strata and weight in this syntax?
>
>and what will be the problem if we use a
>
>simple proc logistic for survey data
>
>
>
>proc surveylogistic;
>  cluster sampunit;
>  strata stratum;
>  model depvar = indvar1 indvar2 indvar3;
>  weight sampwt;
>run;

What will be the problem if you run this using PROC LOGISTIC instead?
Well, your standard errors will be wrong.  Your test statistics will be
wrong.  Even your degrees of freedom will be wrong.  So everything
you report - except your point estimates - will be wrong.  Other than
that...

What will be the difference between cluster, strata and weight
statements?  There's a huge fundamental difference.  It's like the
difference between a MODEL statement and a CLASS statement.
They don't work anything like the other statements.  If you have
a stratified sample, you need to tell the proc about your strata.
If you have a cluster sample, you need to tell the proc about your
clusters.  If you have sample survey weights, then you need to
list them.  If you have a multi-stage sample, then you have to
understand how the SAS survey analysis procs take the design
information in.

HTH,
David
--
David L. Cassell
mathematical statistician
Design Pathways
3115 NW Norwood Pl.
Corvallis OR 97330

_________________________________________________________________
On the road to retirement? Check out MSN Life Events for advice on how to
get there! http://lifeevents.msn.com/category.aspx?cid=Retirement