Logistic regression - "unique profiles"

[saurabh1]

Hello,

I am doing logistic regression in R on a binary dependent variable with only one independent variable. I found the odd ratio as 0.99 for an outcomes. This can be shown in following. Odds ratio is defined as, ratio_odds(H) = Probability(X=H) / (1-Probability(X=H)). As given earlier ratio_odds (H) = 0.99 which implies that the probability (X=H) = 0.497 which is close to 50% probability. This implies that the probability for having a H cases or non H cases 50% under the given condition of independent variable. This does not seem realistic from the data as only ~20% are found as H cases. Please give clarifications and proper explanations of this kind of cases in logistic regression.

Thanks all for your help!

[Reeza]

If you're using R, why are you posting to a SAS forum? Casting a wider net for an answer to your question?

IMO a question like this should be posted on stats.stackexchange.com

[Ksharp]

1)Maybe you missed some important independent variables .
2)Maybe the Scale for independent variable is too small . E.X. the dose of medication :  1ml v.s  100ml . 
   Check PROC LOGISTIC 's  UNIT statement.
3)Maybe there are some none-linear effect between  independent variable and dependent variables.
  Check PROC LOGISTIC 's  EFFECT statement.

4)Why not use Decision Tree or Random Forest in R ?