# Re: st: RE: RE: CIs for the ratio of two adjusted predictions after logistic?

 From Maarten buis To statalist@hsphsun2.harvard.edu Subject Re: st: RE: RE: CIs for the ratio of two adjusted predictions after logistic? Date Wed, 1 Aug 2007 17:31:19 +0100 (BST)

```--- Nick Cox <n.j.cox@durham.ac.uk> wrote:
> I think in this the term
>
> _b[foreign]
>
> should be
>
> _b[foreign] * foreign
>
> and that
>
> _b[foreignXmpg] * mpg
>
> should be
>
> _b[foreignXmpg] * foreignXmpg

No, I am comparing for each value of mpg the predicted probability of
foreign cars with the predicted probabilities of domestic cars. To do
so I create a local `xb_for' which contains the linear predictor
assuming the car is foreign and a local `xb_dom' which contains the
linear predictor assuming that the car is domestic. (actually the
locals contain the commands that will create the linear predictors, but
that is a detail) Consider the code for the linear predictor assuming
that the cars are foreign.

#delim ;
local xb_for "_b[_cons] + _b[foreign] +
_b[mpg]*mpg + _b[foreignXmpg]*mpg" ;
#delim cr
In that case we know that the the variable foreign has value 1, so

_b[foreign]*foreign =
_b[foreign]*1 =
_b[foreign]

similarly:

_b[foreignXmpg]*foreign*mpg =
_b[foreignXmpg]*1*mpg =
_b[foreignXmpg]*mpg

-- Maarten

-----------------------------------------
Maarten L. Buis
Department of Social Research Methodology
Vrije Universiteit Amsterdam
Boelelaan 1081
1081 HV Amsterdam
The Netherlands

Buitenveldertselaan 3 (Metropolitan), room Z434

+31 20 5986715

http://home.fsw.vu.nl/m.buis/
-----------------------------------------

___________________________________________________________
Yahoo! Answers - Got a question? Someone out there knows the answer. Try it
now.