I want to test for the above trend using Proc Catmod. I have 'age', 'indication', 'type of hospital' and 'source of payment' as my 'designs-effects' variable. I have both the frequencies (counts) and the rates for C-Sections in the State of New England for the 6 years years 2000 - 2005.
I need help regarding the following:
1. What variable should I use as the option for the 'weight' statement? The 'count' of C-Sections or their actual 'rate'?
2. Since I'm using the weighted least squares approach, I'll include the option wls in my model statement. What option should I use for the response statement?
3. I intend to treat the variable 'year' as the response effect. Is this the right approach? If not, do you have an alternative suggestion?