[Gretl-users] Re: gretl vs Stata: different WLS estimates for weights=0
On Wed, 20 Jan 2021, Artur Bala wrote:
Le mer. 20 janv. 2021 à 15:05, Sven Schreiber
<svetosch(a)gmx.net> a écrit :
> Am 20.01.2021 um 14:26 schrieb Artur Bala:
>
> I don't know what Stata does there (although I suspect it is documented),
>
Yes, sure. To be frank I haven't researched this yet as I was surprised by
the difference. [...]
If the weight series is all positive, or is a 0/1 dummy, gretl and
stata produce identical results. If it's not a dummy but contains
zeros (as in your example), it seems that gretl is not handling the
number of observations in a fully consistent manner. The parameter
estimates are the same as stata but I think the gretl standard
errors are off, due to not factoring in the reduced number of
effective observations at the right stage of the calculations.
This should be fixed shortly.
Allin