I am investigating the difference in childhood obesity rates between Dudley and Brighton & Hove. These areas have similar deprivation rates but very different childhood obesity rates. I am looking at fast-food outlets, green space, supermarkets, car ownership and education to see if there if a relationship with childhood obesity; this is at ward level. What would be the best way to compare the results between these two areas?
Asked
Active
Viewed 26 times
1
-
Who or what are Dudley and Brighton & Hove? I guess they are towns or some such, but adding a word or two to explain that might make your question easier to understand for those from other countries. – Aug 10 '15 at 08:54
-
1Sorry, Dudley is a town in England and Brighton & Hove is a City. They are both similar sizes and made up of around 20 wards. – Erynn Aug 10 '15 at 09:42
1 Answers
1
I would start with a Poisson rate regression, it is generallt better to model counts than the rates directly. Assuming you have counts on ward level, the response is count of obese children and the relevant population count is represented as an offset. You can find more information in some other posts here:
How is a Poisson rate regression equal to a Poisson regression with corresponding offset term?
Goodness of fit and which model to choose linear regression or Poisson
With Poisson regression there is always the possibility of overdispersion (search this site), so that must be investigated, and maybe some variant such as a quasi-Poisson or negative binomial model should be used.
kjetil b halvorsen
- 77,844