I'm trying to run a difference-in-difference to measure the impact of an intervention on medical costs. My dataset has information on yearly medical costs across a 13 year time frame, and the cost data is all positive but highly negatively skewed.
I tried running the following command:
xtgee cost interventionarea treatedyr*interventionarea i.year family(gamma) link(log) corr(exchangeable) robust
but I get an error saying convergence is not possible. I understand that I should make the model simpler, but that's not possible in this scenario where I only have three coefficients
I've been recommended to use gllamm instead but I'm not sure how to use this for my model. Does anyoen have any suggestions?
I tried running the following command:
xtgee cost interventionarea treatedyr*interventionarea i.year family(gamma) link(log) corr(exchangeable) robust
but I get an error saying convergence is not possible. I understand that I should make the model simpler, but that's not possible in this scenario where I only have three coefficients
I've been recommended to use gllamm instead but I'm not sure how to use this for my model. Does anyoen have any suggestions?
