Date: Mon, 24 Aug 1998 15:58:37 -0700
Reply-To: Elizabeth A Demers <pedemers@LELAND.STANFORD.EDU>
Sender: "SAS(r) Discussion" <SAS-L@UGA.CC.UGA.EDU>
From: Elizabeth A Demers <pedemers@LELAND.STANFORD.EDU>
Subject: probit model
Content-Type: text/plain; charset=US-ASCII
I'm using Probit in SAS for the first time and have
one quick question... I have a dichotomous response
variable where 0's indicate that the event of interest
(response) did not occur and 1's indicate that the event
did occur. It is most intuitive to think about our
research question in terms of whether the independent
variables are associated with the probability of the
event *occuring* rather than with the probability that
the event does not occur. I understand that SAS models
the probability that the lesser response occurs (i.e.,
the 0's in my case).
Question: Is there any way to reverse which probability
SAS estimates with proc probit??
(I know that there's an easy way to do this for logistic regressions.)
Of course I could just recode the 0's & 1's, but I'm
using the same variables and dataset in other aspects
of the study so would like to retain consistency for
ease of moving from one aspect of the study to another
without error or confusion between researchers.
Thanks if anyone can help.