Problems in a dichotomous logistic regression

Home Forums Methodspace discussion Problems in a dichotomous logistic regression

Viewing 2 posts - 1 through 2 (of 2 total)
  • Author
  • #1615

    Hi guys! I am making a dichotomous logistic regression which probably has a problem. There are two variables whose EXP (B) – 95% intervals are for example 0.9 – 333.5, I think that this fact indicates a mulicolineality problem, but I am not sure and, in that case, I don’t know how to manage it. The problematic variable is a dichotomous one, which has been recoded into (0-1) form. I will be grateful if you can explain me about this topic in order to solve the problem or even if you post some resources (articles or books) which could help me! Thanks a lot!

    Jeremy Miles

    It’s not necessarily a problem – that’s your confidence interval. However, it’s a sign that you have quasi-complete separation, that is very few people left in each cell.  Say you have X (binary predictor) and Y (binary outcome) then you have four possible combinations: 0,0; 0,1; 1,0; 1,1. You need a reasonable number of people in each of those cells. If you don’t, you’ll get the sort of results that you have. And as the number of predictors increases, this gets more complex.

Viewing 2 posts - 1 through 2 (of 2 total)
  • You must be logged in to reply to this topic.