Interpretation of Goodness of fit tests in a Poisson regression

[Francoise Ren]

Hello,
I am studying mortality rates , and try to find out which variables can best explain the rates
I have run a Poisson regression, then looked at the goodness of fit of my model.
I do not understand the results of the gof tests (deviance and Poisson) : they both give extreme results , 0 and 1;
Moreover, it looks like they give opposite results, so I don't know how to conclude about my model.
Could you help me interpret those results ?
Did I make something wrong ? thanks, Françoise


"poisson _d curage i.gewest ib4.educat4 i.empl4 ib4.newcomf2 if sex==1&educat4!=.&empl4 !=.&newcomf2!=.&owner!=.&isol!=.,e(y) irr
estat gof "

RESULTS:

estat gof

Deviance goodness-of-fit = 61023.65
Prob > chi2(443788) = 1.0000

Pearson goodness-of-fit = 3062899
Prob > chi2(443788) = 0.0000

[Marcos Almeida]

Hello Fracoise,

Welcome to the Stata Forum.

Please provide commands and results under CODE delimiters as recommended in the FAQ.

The interpretation of the gof estimations are explained in the example 1, here: http://www.stata.com/manuals13/rpois...estimation.pdf

As you may see, we shouldn't expect opposite - in your case, extreme! - values.

Best,

Marcos

[Francoise Ren]

Hello Marcos, thanks for advising me for the way of writing the code.
I realize that my question was not clear: I know the theory and already visited the Stata website on this topic; my problem is that my results are strange. I was wondering what was wrong and could cause such unexpected results? So I shall reformulate all this in a ne post. Thanks anyway, Françoise