Poisson Regression with Probability Weights and Clustered Standard Errors

[Trevor Andrew]

When I estimate a Poisson regression with probability weights or clustered standard errors the r-squared is not reported. Can please someone explain why and what is the work around?

[Carlo Lazzaro]

Trevor:
because -poisson- returns -e(chi2)-.
If what above is not the answer you were looking for, -help j_robustsingular- may be helpful.

[Trevor Andrew]

Thanks Carlo, the help documentation is a great reference! If I read it correctly there is no work around to get pseudo R-squared if I use probability weights or clustered standard errors in a Poisson regression? Negative Binomial regression does report pseudo R-squared if I use probability weights or clustered standard errors.

[Carlo Lazzaro]

Trevor:
usually, there are two main reasons why Stata does nit report a given statistic:
1) under a given set of conditions, that statistic is too complex to be caluculated;
2) the statistic is not sufficient and/or suffer from other methidological drawbacks that make it unreliable/misleading.