Date:         Fri, 16 Dec 2005 20:03:16 -0800
Reply-To:     David L Cassell <davidlcassell@MSN.COM>
Sender:       "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From:         David L Cassell <davidlcassell@MSN.COM>
Subject:      Re: Interpreting an intercept
In-Reply-To:  <200512162305.jBGL2d2p014958@mailgw.cc.uga.edu>
Content-Type: text/plain; format=flowed

deepstar@GMAIL.COM wrote:
>I have been interpreting intercepts in the past without problems, but
>now i'm facing a multiple regression analysis with columns like:
>
>                      b
>Woman:       -6.6
>Caucasian:    8.5
>Experienced: 11.4
>
>Intercept/constant B = 5.5
>
>The numbers show the impact the three top variables have on a dependent
>variable (salary). How am I supposed to interpret the intercept here?

I don't know whether the above accurately reflects what you are doing, or
not.
After all, you say your real results are columns LIKE what you show us.  But
I'll
take a guess, based on what you showed us.

So you have a dummy variable WOMAN which is 1 for women and 0 otherwise.
You have a dummy variable CAUCASIAN which is 1 for gringos and 0 otherwise.
You have a dummy variable EXPERIENCED which is 1 for k or more years
working,
and 0 otherwise.  And you have an intercept.  And nothing else.  (This seems
like a really over-simplified model, so I hope you are checking your
regression
diagnostics and making sure your real problem has a reasonable model.)

This means that the intercept is not just an intercept.  It is the value of
your
response when WOMAN=CAUCASIAN=EXPERIENCED=0 .  So if you have
inexperienced non-white male workers, this is your mean salary estimate for
them, according to your model.

HTH,
David
--
David L. Cassell
mathematical statistician
Design Pathways
3115 NW Norwood Pl.
Corvallis OR 97330

_________________________________________________________________
On the road to retirement? Check out MSN Life Events for advice on how to
get there! http://lifeevents.msn.com/category.aspx?cid=Retirement