Poisson command on panel data

[Olive Bat]

when i run the following commnad
xtpqml lexp_food year mig_stats_16 mig_stats_yr sex sector edu_hh employ marital_st HSIZE, fe i(HHID)
note: you are responsible for interpretation of non-count dep. variable

the result reveals "note: you are responsible for interpretation of non-count dep. variable".. why does this happen

[Joao Santos Silva]

Dear Olive Bat

Poisson regression was originally proposed for count data, that is why Stata warns you if the dependent variable is not a count. It turns out that it is fine to use Poisson regression for data that are not counts, so you can ignore the warning.

On a side note, please make sure the dependent variable in the Poisson regression is not logged.

Best wishes,

Joao

[Carlo Lazzaro]

Olive:
does -lexp_food- takes on >0 integers only?
As a seemingly annoying aside, please tell us when you use a community-contributed Stata command (FAQ explain good reasons for that). Thanks.

[Olive Bat]

Carlo Lazzaro ...yes sir.. it take intergers/values greater than zero

[Olive Bat]

Joao Santos Silva wll keep that in mind Sir.. thanks for the clarification..

[Carlo Lazzaro]

Olive:
"Sir" makes everything too official...Carlo is enough! Thanks.""

[Richard Williams]

Also, Bill Gould has a good post about using Poisson with non-count variables:

https://blog.stata.com/2011/08/22/us...tell-a-friend/