[Gretl-devel] Strange bug in count data models

[Allin Cottrell]

On Tue, 1 Apr 2008, Riccardo (Jack) Lucchetti wrote:

> it seems we have a bizarre bug in our Poisson estimation code: if the
> dependent variable is called "d_avril" we get bogus estimates. 

Could you sed a test case, please?  Just calling the dependent 
"d_avril" does not cause the problem, as you can see by renaming 
the dependent variable in greene19_3.inp as d_avril.

open mccullagh.gdt
genr d_avril = Acc 
poisson d_avril const TB TC TD TE T6569 T7074 T7579 O7579 ; Months
poisson d_avril const T6569 T7074 T7579 O7579 ; Months
poisson d_avril const TB TC TD TE O7579 ; Months
poisson d_avril const TB TC TD TE O7579 ; Months

This all works fine (here, at any rate).

Allin.