search

sergiocorreia / ppmlhdfe

Poisson pseudo-likelihood regression with multiple levels of fixed effects
http://scorreia.com/software/ppmlhdfe/
MIT License
64 stars 13 forks source link

ppmlhdfe aweights #2

Closed CorinnaHa closed 5 years ago

CorinnaHa commented 5 years ago

Hi Sergio,

it would be great if you could program the command "aweights" for ppmlhdfe. The commands "fweight" and "pweight" work well.

Thanks! Corinna

sergiocorreia commented 5 years ago

Hi Corinna,

The reason for not implementing aweights is that ppmlhdfe needs to use pweights to control for the fact that you might be modelling a "pseudo" poisson model (ie a model where the variance might not follow what you expect from a poisson distribution).

We can combine that with frequency weights, but because pweights is just robust aweights, it was not something you can do in pseudo models.

Are you thinking about running a (non pseudo) poisson model, or something else?

CorinnaHa commented 5 years ago

Thanks for your fast response and the explanation! I am thinking about running a pseudo poisson model using the number of diagnoses per county as outcome variable. I intended to weight observations by the number of individuals living in each county using the command aweights.

sergiocorreia commented 5 years ago

I don't think aweights is valid in this case, instead of just using pweights

CorinnaHa commented 5 years ago

Ok, thanks for your help!