Date: Tue, 19 Aug 2008 21:35:13 -0400
Reply-To: Peter Flom <firstname.lastname@example.org>
Sender: "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From: Peter Flom <peterflomconsulting@MINDSPRING.COM>
Subject: Re: Logistic regression math question, Wald v. likelihood ratio
Content-Type: text/plain; charset=UTF-8
Bill Behrens <mngorgeman@GMAIL.COM> wrote
> What's your sample size? In general for small samples LR and Score give
>better results. Wald test relies heavily on normality assumption and in
>small samples this matters. I am not sure if this changes much with
>logistic regression but knowing that it's a nonlinear transformation of
>the linear probability model I wouldn't be surprized if the test results
>are qualitatively similar for small samples.
It was a very small sample ---- I forget exactly right now, but under 50.
Peter L. Flom, PhD
www DOT peterflom DOT com