Re: from odds to probs
From: Alain Verghote (alainverghote_at_yahoo.fr)
Date: 01/18/05
- Next message: Tim: "unmet condition GLM"
- Previous message: Richard Ulrich: "Re: New Pearson's chi-square test?"
- In reply to: Doug: "from odds to probs"
- Messages sorted by: [ date ] [ thread ]
Date: Tue, 18 Jan 2005 13:44:30 +0000 (UTC)
On 16 Jan 2005 22:30:15 -0800, Doug wrote:
>Hi,
>
>This simple enquiry is one that I haven't solved this morning and is
>based on the odds ratio for logistic regression and the associated
>probabilities.
>
>The odds ration equation is O=P/(1-P) (odds=prob/1-prob) of event.
>However in Paul Allison's logiistical regression book (page 11) he
>also has P=O/1+O. While I know that this is true and is easily
>checked with real data (eg 0.2=0.25/1+0.25) and (0.25=0.2/0.8), I
>haven't done this type of maths for a while.
>
>Could someone remind me of the maths steps in getting from O=P/1-P to
>P=O/1+O.
>
>It's monday - ok!
>
>thanks
>
>Doug
Dear Doug,
Just consider function f(x) or f = x/(1+x) (1) ,
To compute the inverse from(1) express x with f;
you obtain: (1+x)*f = x ... and x = f/(1-f) .
Voilą,
Alain.
- Next message: Tim: "unmet condition GLM"
- Previous message: Richard Ulrich: "Re: New Pearson's chi-square test?"
- In reply to: Doug: "from odds to probs"
- Messages sorted by: [ date ] [ thread ]
Relevant Pages
|
