The conclusion section says that "the varaince is higher than predicted by the binomial or poisson model for the range of ages considered. " Please explain
Hi rinishj28, We are looking at graduated rates, which come from fitting a binomial or Poisson) against the original crude rates. We want the graduated rates to be close to the original but they're never going to be exact. How close is close enough, is effectively defined by the denominator in the test statistic (which either stems from a binomial or Poisson, depending on what has been fitted). So, it might be that the graduated rates are a load of rubbish - fine Or it might be that the graduated rates are actually close enough, it's just that we haven't been generous enough in how close is close enough. ie that the true variance is just naturally greater than that allowed by the Poisson or binomial model. Hope this helps, John
