> -----Original Message----- > From: SAS(r) Discussion [mailto:SAS-L@LISTSERV.UGA.EDU] On Behalf Of Irene > Zweimüller > Sent: Friday, January 12, 2007 9:22 AM > To: SAS-L@LISTSERV.UGA.EDU > Subject: logistic regression > > Dear all, > > > > I´ve just calculated a binary logistic regression. > > Under the heading Association of Predicted Probabilities and Observed > Responses > > I found: 80,1% Percent Concordant and I thought, that this means 80,1% of > the observations have been > > Classified correctly with the logistic regression. > > However, when I looked at the probabilities for group membership and the > group membership > > according to the logistic regression I found that only 65% of the > observations were grouped correctly. > > Does anybody have an explanation?? > > > > Kind regards > > Irene Zweimueller > >

Irene,

As you have discovered, Percent concordant is not percent correctly classified. For percent concordant, one compares each and every CASE (what you are predicting) to each and every nonCASE. So if you have 100 CASEs and 100 nonCASEs you will have 10000 comparisons. Percent concordant is the percentage of these comparisons in which the predicted probability of being a CASE is greater for the CASE than for the nonCASE.

Hope this is helpful,

Dan

Daniel J. Nordlund Research and Data Analysis Washington State Department of Social and Health Services Olympia, WA 98504-5204