1

I am modeling rates of people given drug $X$ or drug $Y$, working out probability of getting drug.

The drug $X$ and drug $Y$ points have been matched on exposure time (person years) to specific drug.

In this case, is there a need to include an offset term or should it now be treated as count data?

Greenparker
  • 14,131
  • 3
  • 36
  • 80
DRSP
  • 11
  • 1
  • http://stats.stackexchange.com/questions/11182/when-to-use-an-offset-in-a-poisson-regression/11183#11183 – ocram Apr 28 '16 at 11:06
  • 1
    Some more detail would be helpful. Is a given person possibly given drug X repeatedly so you have a count per person? Could each person possibly be given drug X and drug Y so you have a comparison within person? – mdewey Apr 28 '16 at 11:22

0 Answers0