You can compute marginal effects at the mean or median of the explanatory variables by appending observations to the end of the input data set and refitting the model. In the new observations, set the explanatory variables to their means or medians, and set the response variable to missing. Refit the model and use the MARGINAL option in the OUTPUT statement to obtain marginal effects for these new observations. This approach is illustrated in this note and in this web example. The general process of scoring new observations in this way is described and illustrated in more detail in this note.