Dear Essoult,
You can use a multinomial logit model. All main
statistical softwares can run it.
Good luck
--- Esoult Hanratty <[log in to unmask]> wrote:
> Hello,
>
> I have a study where I know the genetype of each
> patient (AA, AB, BB). I
> have been using logistic regression in order to
> model the risk factors
> associated with each genetype (for example AA versus
> AB or BB). This means
> that I have 3 logistic regression models, is this an
> accurate analysis to
> do? When you have a variable that has three outcomes
> is it OK to use
> logistic regression to predict one of the outcomes?
>
> I realise that there is overlap in doing the 3
> logistic regressions - so in
> that sense it might not ba appropriate.
>
> Any advice would be appreciate, logistic regression
> is fairly new to me.
>
> Regards,
>
> Esoult
>
>
_________________________________________________________________
> Find just what you're after on your PC with Windows
> Desktop Search - FREE!
> http://desktop.msn.ie/
>
__________________________________________________
Do You Yahoo!?
Tired of spam? Yahoo! Mail has the best spam protection around
http://mail.yahoo.com
|