I have to recreate a model that was build in Stata using R. I want to calculate marginal effects after using a ordered probit/logit model. I have a dependent (ordinal) variable with 5 outcomes and several independent variables.
I tried "MASS" and "erer" as well as "oglmx"-package and all of them provide me with marginal effects for every outcome of my DV. Like:
x <- ocME(m1, x.mean=TRUE)
x
effect.Str agree effect.Agree effect.Disag effect.Str disag
x1 -0.198 -0.047 0.076 0.169
x2 -0.050 -0.012 0.019 0.043
x3 -0.018 -0.004 0.007 0.016
However, neither package seems to have an marginal effect function that simply gives you one single marginal effect value for each independent variable. I want to calculate a single Marginal Effects at the Means for every independent variable - see in Stata:
I think the code in stata is: . margins, dydx(*) predict(outcome(1))
Almost the same question as: Finding Marginal Effects of Multinomial Ordered Probit/Logit Regression in R . But I think there was a misunderstanding what values he wanted, that's why there wasn't a clear answer.
I hope I made clear what kind of marginal effect I want. You would be my lifesavers if someone knows a good package or how to calculate them manually. Thank you all in advance!
