Hi there,

I have a categorical variable 'Kickername', which gives the name of an NFL field goal kicker. I have used the code:

encode Kickername, gen(Kickername2)

to destring the variable. I then used i.Kickername2 in a logit regression on if a kick was successful or not to get a dummy variable for each kicker (a 'kicker effect')

However, when I try to get marginal effects from this using the code:

margins, dydx(*) atmean

It says that the marginal effects for all of the kicker dummy variables are not estimable. It generates marginal effects for all other variables in my regression with no issue.

I tried generating the dummy variables for each kicker manually, but I have been told that 'margins' does not work without using factor variable notation - is this correct?

If so, any advice as to why marginal effect for i.Kickername2 would be greatly appreciated - Thanks!