[Gretl-users] Re: Marginal effects calculation - RE probit & FE logit
Am 03.02.2021 um 12:01 schrieb martina.marseglia2(a)studio.unibo.it:
I am running FE logit (function package of Lucchetti) and RE probit
but there is no option for calculating slopes and different kinds of marginal effects. I
downloaded the function packages a_eff of Komashko e lp-mfx of Cottrell but it seems there
is no possibility to run them on FE logit and RE probit.
I need marginal effects for evaluating my regression, how can I calculate them?
There's also the margeff_el package, also by Oleh (Komashko). I _think_
it covers those kinds of problems in principle. The documentation might
be a bit difficult to read, but it does exist. Of course for such
non-linear models you have to decide at what level of the variables you
want to evaluate the slopes/effects.
I also know that in the DPB package there is a function for computing
average partial effects for a dynamic panel probit model. Maybe the
package authors could adapt that code to the static RE probit case?
cheers
sven