Date: Thu, 11 Dec 2003 10:29:59 -0500
Reply-To: "Will Bailey [Statman]" <statmanz@earthlink.net>
Sender: "SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>
From: "Will Bailey [Statman]" <statmanz@earthlink.net>
Subject: Re: weighted data and regression models
In-Reply-To: <9A782E8087FED411868000508B69311C88BBEE@CCHEXCHANGE>
Content-Type: text/plain; charset="US-ASCII"
You can't using 'standard' tools such as SPSS base but must move over to
those that handle complex sampling, e.g., SPSS' new Complex Sampling module
[however, does not do regression in the V12 version] or www.westat.com
WMB
Statistical Services
=========================================
mailto:willbaileyz@earthlink.net
http://home.earthlink.net/~statmanz/
=========================================
Virus Scan Notice: This email is certified to be virus free.
-----Original Message-----
From: SPSSX(r) Discussion [mailto:SPSSX-L@LISTSERV.UGA.EDU] On Behalf Of
Schaible-Brandon,Sonia
Sent: Thursday, December 11, 2003 10:25 AM
To: SPSSX-L@LISTSERV.UGA.EDU
Subject: weighted data and regression models
If SPSS weights to a population size, how does one adjust for the "error" in
regression by the artifical increase of the degrees of freedom? It would
seem that the standard error, particularly if some cases are weighted much
more than others would be biased. How would I know if it is it necessary?
Could one run bootstrap samples on the weighted population?
Sonia Schaible-Brandon
Colorado Commission on Higher Education
1380 Lawrence Street Suite 1200
Denver, CO 80204
phone: (303)866-2723
fax: (303)866-4266
email: sonia.schaible-brandon@state.co.us
