Using marginal effects after cprobit appropriate?

[Natan Sklair]

Hi there,

Is reporting marginal effects atmeans i.e." margins, dydx(*)" after a panel logit(clogit) with fixed effects appropriate?

I have seen different opinions online - that reporting margins is incorrect for this type of analysis.

Thank you!

[Clyde Schechter]

The reporting of predictive margins after -xtlogit,fe- or -clogit- is inappropriate. (By that I mean the result of -margins- without a -dydx()- option specified) as those parameters are not identifiable in a fixed effects logit model. But the marginal effects, that is, those with the -dydx()- option are identified: they are, after all, simply linear combinations of the regression coefficients themselves and do not involve either the fixed effects or constant terms. Stata will not report them unless you specify the -noestimcheck- option, but it is reasonable to do so in this case.