Dear Grelt list members,
I run a random effects probit regression and was wondering, why the output displays
p-values, though the -p-values option was not selected. When I estimate the model as a
pooled (panel) probit, then Gretl displays the slopes, which corresponds to the command
description.
Is there another option (accessor variable, etc.) which enables one to retrieve the
average marginal effects - averaged over each entity and instead of evaluating the effects
at the average of the independent variables? I ask because I am using some binary
independent variables, and evaluating the marginal effects at some value in between would
not make much sense for these variables.
FYI the LR test that tests Ho: rho=0 is rejected with a p-value 0.000173
Thanks in advance.
Jan
Show replies by date