Date:         Fri, 29 Feb 2008 15:14:55 -0600
Reply-To:     Mary <mlhoward@avalon.net>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         Mary <mlhoward@AVALON.NET>
Subject:      Re: Bivariate and multivariate results
Comments: To: Peter Flom <peterflomconsulting@mindspring.com>
Content-Type: text/plain; charset="UTF-8"

If you are doing logistic regression couldn't you present the significance level of each regression run and possibly the odds ratios, one line per regression run?

-Mary
  ----- Original Message ----- 
  From: Peter Flom 
  To: SAS-L@LISTSERV.UGA.EDU 
  Sent: Friday, February 29, 2008 2:33 PM
  Subject: Bivariate and multivariate results


  Hi

  I've got a client who is running a lot of multiple regressions. She also wants to present the bivariate results.  I've run this as a PROC CORR with the WITH statement, for ease of programming and compact output.  I think this makes sense.
  But I've not seen results presented this way (i.e. a table of correlations, and then a multiple regression).

  Now, she wants to do a similar regression on a dichotomous DV, which, of course, calls for logistic regression.  I could recode the DV as 0 1, and again present the correlations.... again, this gives compact output, and a nice simple program.  It's meaningful, since the DV is dichotomous.  But will it fly in a journal?

  Or am I missing something more fundamental?

  TIA as always

  Peter

  Statistical Consultant
  www DOT peterflom DOT com