.999 signifiance query

Question

I am getting the following result for one of my variables after running a logistic regression.

Basically I have a set of binary dependent and one independent variable. One of the dependent variables is never 0 when the independent variable is 1. I believe that is why I am getting the above result. How would you interpret this? The .999 indicates that it's not significant but surely it should be? Also, what would you say about the odds ratio?

EDITED: Yes, my bad. I have one DV and about 10 IVs. All of the other IVs display normal ORs and 4 have significant p values. The 'endoc' one I included in the screenshot above is the only 'outlier'. Here is the rest of the regression table: . and this is part of my data

I am basically writing a report for this, so I just need to know what to say about this occurence?

Answer 1

This problem is called separation and some suggestions for how to deal with it are here.

Answer 2

This is clearly problematic output. But it is also problematic text. In a logistic regression you should have one dependent variable and one or more independent variables, not the other way 'round.

Assuming this was a typo, and you really do have just one DV, then this looks like colinearity. But if, at step 1, there is only one IV, you can't have colinearity, which requires more IVs.

If you have one IV and one DV and the IV is never 0 when the IV is 1 (or something like that, it's hard to tell) then you would get an OR of 0 or infinity; you didn't get infinity, but that could be some computational error. You shouldn't do a regression like this.

Please clear up the above confusions. Showing your code and a sample of your data would help.