Poisson regression - nocons

[Lucia Carra]

Dear all

I am conducting a mirror image study on days of admission to hospital before and after a certain treatment. I am using negative binomial regression but I'm not sure whether I should use nocons or not. How can I understand it?

Many thanks to everyone

Lucia

[Phil Bromiley]

You'll increase your chances of a useful answer by following the FAQ on asking questions - provide Stata code in code delimiters, readable Stata output, and sample data using dataex.

If you search this listserve, you'll find many interesting comments on negative binomial and poisson estimators. Most of us automatically include a constant. While it may not work this way exactly in nbreg, in regression not including a constant forces your predicted y to equal zero when all the x's equal zero - that is, the fitted line must equal zero when all the x's equal zero - which is not an assumption most of us would want to impose.