Upon further reflection, it seems to me that the original poster may not have a categorical variable with 5 levels, but rather four categorical variables, each with 2 levels. This has not been brought up for discussion here. But in the original post, it was said that communication he/she had received on this topic indicated that the Wald test would have 2 df. Thus, it seems likely that the intention was to construct tests such as the following:

d1: test d1=0, xd1=0; d2: test d2=0, xd2=0; d3: test d3=0, xd3=0; d4: test d4=0, xd4=0;

Now, these tests only make sense if the four dummy variables represent binary conditions to begin with. These tests do not make any sense if the dummy variables represent four of five levels of a categorical variable.

Dale

--------------------------------------- Dale McLerran Fred Hutchinson Cancer Research Center mailto: dmclerra@NO_SPAMfhcrc.org Ph: (206) 667-2926 Fax: (206) 667-5977 ---------------------------------------

--- On Thu, 5/29/08, Robin R High <rhigh@UNMC.EDU> wrote:

> From: Robin R High <rhigh@UNMC.EDU> > Subject: Re: Wald test > To: SAS-L@LISTSERV.UGA.EDU > Date: Thursday, May 29, 2008, 1:06 PM > PA, > > If you dummy coded the data (both the class var d and its > interaction as > GLM does for you, based on Dale's code) you can make > the same test in PROC > REG with a TEST statement: > > proc reg; > model y = d1 d2 d3 d4 x xd1 xd2 xd3 xd4 ; > TEST d1=0, d2=0, d3=0, d4=0, xd1=0, xd2=0, xd3=0, xd4=0; > run; quit; > > Robin High > UNMC > > > > > alves <alves.paulo@GMAIL.COM> > Sent by: "SAS(r) Discussion" > <SAS-L@LISTSERV.UGA.EDU> > 05/29/2008 12:10 PM > Please respond to > alves <alves.paulo@GMAIL.COM> > > > To > SAS-L@LISTSERV.UGA.EDU > cc > > Subject > Re: Wald test > > > > > > > Hi Dale, > > Thanks for your prompt answer. Not sure if it changes > anything in your > answer, but I am using PROC REG for my regressions. the > number we use > 5 dummies. > > thanks again, > > PA