Date: Wed, 16 Sep 2009 13:00:27 -0700
Reply-To: Steve Denham <stevedrd@YAHOO.COM>
Sender: "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From: Steve Denham <stevedrd@YAHOO.COM>
Subject: Re: Draize scoring
In-Reply-To: <705bcd670909161058r514d950ak3615d55d6b997bd3@mail.gmail.com>
Content-Type: text/plain; charset=iso-8859-1
You might consider an exact Mantel-Haenszel test (see the PROC FREQ documentation), if you had just one or two timepoints. If you are really interested in the time by treatment interaction (or the sponsor is {or should be}), then I'm out of ideas, since the multinomial distributions in GENMOD generally don't do well with repeated measures. I'd go for a graphical exploration first to see if there is a critical date when separation occurs, and then back to the exact MH.
Let us know what you decide. It is important.
Steve Denham
Associate Director, Biostatistics
MPI Research, Inc.
----- Original Message ----
From: joey m <sasuser.joey.m@GMAIL.COM>
To: SAS-L@LISTSERV.UGA.EDU
Sent: Wednesday, September 16, 2009 1:58:39 PM
Subject: Re: Draize scoring
Hi Paul:
Thanks a lot. Sorry for my delay to replay. It was caused by a virus...not
in my computer...in my body. I know this questions is more statistical that
SAS, but I have been reading some papers to get some light, but no luck so
far. Now, healthy again, I will restart my reading.
Not only the number of categories is small, also,the number of subjects.
They will be (it is a research in preparation) 25-30 subjects. They will be
assigned to 4 or 5 treatments (more or less 3 animals per group, for each
sex), and scoring will be done daily, during a period of 14 days.
Thanks, I appreciate your help.
J.
On Wed, Sep 9, 2009 at 5:00 PM, Swank, Paul R <Paul.R.Swank@uth.tmc.edu>wrote:
> Given the small number of categories, I don't think Wilcoxon would be best
> because of the number of ties. How many subjects do you have. A generalized
> linear model (multinomial) or logistic (event/trials) may be a better way to
> go.
>
> Dr. Paul R. Swank,
> Professor and Director of Research
> Children's Learning Institute
> University of Texas Health Science Center-Houston
>
>
> -----Original Message-----
> From: SAS(r) Discussion [mailto:SAS-L@LISTSERV.UGA.EDU] On Behalf Of joey
> m
> Sent: Tuesday, September 08, 2009 9:52 AM
> To: SAS-L@LISTSERV.UGA.EDU
> Subject: Draize scoring
>
> Hi all:
>
> Does someone know a paper (or any SAS reference) to check how to fit a
> non-parametric analysis for Draize Scoring (a subjective evaluation of
> sking
> irritation; i.e. edema, erithema, etc; scale goes from 0 to 4; being 0 no
> response, and 4 severe response; all scored at several time points on the
> same animals). I would like to compare several groups against a control at
> each time point.
>
> I believe a Wilcoxon test must do, but I will appreciate any suggestions.
>
> Nice day to all,
>
> Joey.
>
| 
