LISTSERV at the University of Georgia
Menubar Imagemap
Home Browse Manage Request Manuals Register
Previous messageNext messagePrevious in topicNext in topicPrevious by same authorNext by same authorPrevious page (June 2000)Back to main SPSSX-L pageJoin or leave SPSSX-L (or change settings)ReplyPost a new messageSearchProportional fontNon-proportional font
Date:   Wed, 28 Jun 2000 22:29:00 -0400
Reply-To:   Muyiwa Oladosu <oladosu@EXCELONLINE.COM>
Sender:   "SPSSX(r) Discussion" <SPSSX-L@LISTSERV.UGA.EDU>
From:   Muyiwa Oladosu <oladosu@EXCELONLINE.COM>
Subject:   Re: Weight Determination (fwd)
Comments:   To: Tim Dunsworth <Tim.Dunsworth@METROSTATE.EDU>
Content-Type:   text/plain; charset="iso-8859-1"

Thanks to everyone who spared some time to enlighten me on weighting issues. This list serve is undoubtedly a rich reservoir of knowledge. I now have enough info to assist me with weighting. Thanks once more.

Muyiwa. ----------------------------------------------------- Click here for Free Video!! http://www.gohip.com/freevideo/

-----Original Message----- From: Tim Dunsworth <Tim.Dunsworth@METROSTATE.EDU> Newsgroups: bit.listserv.spssx-l To: SPSSX-L@LISTSERV.UGA.EDU <SPSSX-L@LISTSERV.UGA.EDU> Date: Wednesday, June 28, 2000 10:54 AM Subject: Re: Weight Determination (fwd)

>Yes, you're right Curt - it was a geezer lapse on my part that I didn't catch because my incorrectly computed weights still managed to preserve the total sample size (125 + 375 = 500, while the correct numbers are 100 + 400 = 500). Thanks for catching the error. > >>>> curtyboy@worldspy.net 06/27 6:39 PM >>> >see below > >> >> There are several types of weights in common use. Hector's reply uses one >definition that has the advantage that all statistics (eg. cumulative number >of cases up to some observed value of a numeric variable) are direct >estimates of the corresponding population values, but it has the >disadvantage of increasing the apparent sample size and thus affecting >significance values. >> >> If you want to preserve the same total sample size, calculate the weights >as the population proportion of each group divided by the sample proportion >of that same group. For urban dwellers that would be: >> W = (1000/4000) / (250/500) = 0.5 > >Tim: > >I'm trying to understand your recommendation. How do you define population >proportion? In this case, I would think that population proportion = >(1,000/5,000), and therefore the weight for urban dwellers would be: >(1000/5000)/ (250/500) = .4. > >And the weight for rural dwellers would be (4,000/5,000)/(250/500) = 1.6. > >Am I on or off target? > >Curt Dommeyer


Back to: Top of message | Previous page | Main SPSSX-L page