I am performing logistic regression and would like some help with interpretation. My data was collected at 2 time points; baseline and follow-up.

My outcome variable is a psychosocial functioning variable (as an example working or not working) at follow-up.

In step 1 I entered the functioning variable at baseline. In step 2 and 3 I entered a number of covariates e.g sex, age, SES, psychopathology variables (dichotomous).

In step 4 I entered a continuous variable which is the BPD difference score and it is dimensional.

So my problem is with interpretation. If I find that step 4 was significant and BPD difference score significantly predicted "working at time 2". If this was a linear regression I could say that there was an improvement in working as working at baseline was controlled for. But I am unsure of how to interpret it when I control for the dichotomous variable in step 1 (working at baseline) in a logistic regression.

Does that make sense?

I have done the analysis in SPSS.

Thank you for any responses and I apologise for my inexperience and possible simple question, I am trying my hardest to understand it but I cannot manage to figure it out without help.


1 Answer 1


To translate from the OLS case to the logistic case, you could simply change "improvement in working" to "improvement in the odds of working".

There are a variety of ways to model pre- post- designs, but if you decide to use this one, that's the way to change it.

  • $\begingroup$ Thank you Peter. I much appreciate it and the simplicity of the comment makes perfect sense. $\endgroup$
    – Sarah
    Commented Jul 23, 2013 at 10:06

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.