Given a regression model, I understand that
Code:
margins, dydx(var1)
gives us the average marginal effect of var1 on the dependent variable.

However, Stata gives the option of more than one variable with dydx (e.g. dydx(*)). What would be the interpretation of this? Is it the average effect on the dependent variable due to a small change in all regressors? Something like a total differential?