1

My data takes the following form: Outcome variable = no of suicides (count data) Explanatory variables = Age (grouped into young, middle and old), Gender, Country, Year I am running a Poisson regression. I have been told that I need to take the population size of each country into account, in order to be able to compare no of suicides in different countries. I have now included population in my dataset, and in my regression. What do I do? - How do I interpret the coefficient of Population, or should I ignore this? - Do I need to do anything else? Thank you in advance!

Alice Ilsley
  • 11
  • 1
  • 3
    http://stats.stackexchange.com/questions/11182/when-to-use-an-offset-in-a-poisson-regression – Glen Mar 29 '17 at 22:35
  • This could be a duplicate as Glen has suggested. – Michael R. Chernick Mar 29 '17 at 22:45
  • If you can get the population size of each group (i.e. the population of young females in Belgium in 1992), then binomial regression might be more appropriate. – Jacob Socolar Mar 30 '17 at 04:03
  • Maybe I am missing something but I here are my thoughts. Right now you are comparing the exact number of suicides among countries. I guess there are more people in China than in Liechtenstein so there will be more suicides in China. Hence, you should compare the relative number I would think. So divide your suicides by population and use that number to create a model. However, you won't we be dealing with count data anymore in that case. – Robbie Mar 30 '17 at 05:19

0 Answers0