Marginal effects after logistic

I have all binary/categorical variables, and I am trying to decide between mfx/margins to obtain marginal effects in Stata. They seem to provide different results, but I'm unclear as to the best one to choose and why. Any advice would be appreciated, thanks!


Less is more. Stay pure. Stay poor.
marginal effects. i came across a paper by Matt Bogard discussing MFX. it didnt address cat vars but was interesting. it was titled something like comparing odds ratios and marginal effects in logistic and linear model with r and sas, staff paper.