5

My outcomes is binary variable. I would like to perform logistic regression model to estimate the treatment effect of Xs. However, the follow up time was longer for outcomes=1 and shorter when outcomes=0. The probability of the outcome can be affected by the follow up time, the longer a subject was followed up, the more chance we observe a outcome=1. I considered including the follow up time as an offset, like poisson regression model. Is it reasonable? Otherwise, how can I control the difference in follow up between the two groups.

kjetil b halvorsen
  • 63,378
  • 26
  • 142
  • 467
Jessie
  • 51
  • 2
  • Might be relevant: http://stats.stackexchange.com/questions/25415/using-offset-in-binomial-model-to-account-for-increased-numbers-of-patients http://stats.stackexchange.com/questions/155575/questionable-diagnostics-for-a-binary-logistic-model/155716#155716 http://stats.stackexchange.com/questions/66792/binary-models-probit-and-logit-with-a-logarithmic-offset/115106#115106 – kjetil b halvorsen Mar 17 '17 at 19:46
  • This sounds like you may potentially have censoring. What does 1 and 0 represent? – Glen_b Aug 13 '17 at 02:05
  • I think this question should be in StackOverflow, this post answers to your question: https://stackoverflow.com/questions/13237940/offset-not-working-in-binomial-glm Cheers – Nico Coallier Mar 30 '16 at 21:09
  • 2
    No, this post belongs here! – kjetil b halvorsen Mar 17 '17 at 19:46

0 Answers0