Re: [Gretl-users] Heckit rho
On Tue, 7 Apr 2009, Riccardo (Jack) Lucchetti wrote:
On Tue, 7 Apr 2009, Starr, Brian wrote:
>
> At the bottom of the Heckit output is a 'rho' metric. Does
> anyone know what it is measuring?
That's the estimated correlation coefficient between the
disturbance term of the main equation and the one of the
selection equation. AFAIK, calling it "rho" is fairly standard,
but I may be wrong. Do you find it too cryptic?
I think maybe that should be mentioned in section 22.6 of the
User's Guide, where heckit is discussed. There we talk of a
lambda but no rho.
Allin Cottrell