Logistic reg w/ ratio as predictor

hlsmith

Omega Contributor
#1
Is there anything I need to know when using a ratio as an independent variable in the model.

Is it better to use log(ratio) ?

Not an easy web search, main hits come up as odds ratio and likelihood ratio.

So what I am interested in is binary variable predicted by ratio!
 

hlsmith

Omega Contributor
#2
So the web told me so far that using ratio may be OK if its value tends to be centrally locate and not near bounds. I also saw a recommendation of breaking it into its components as individual predictors. That wont work for me in that the ratio is the measure I am interested in.

Any other suggestion or info I need to know?