[Limdep Nlogit List] Marginal effects for dummy variables in multinomial logit

[Martin Petrick]

Dear List,

does anyone know whether marginal effects for dummy variables in the 
multinomial logit model are calculated differently from the marginal 
effects for continuous variables?

There is an important difference in the binary choice models (described 
in NLOGIT4 manual p. N6-10). No such difference is mentioned in the 
multinomial logit (cf. p. N9-11), although the example given on p. N9-12 
apparently contains a dummy variable (HHKIDS).

Thanks for your consideration,

Martin