[Limdep Nlogit List] NLogit probabilities for ranked data

[contactemt]

Hi,

I understand for a discrete choice the "probs" function returns the probability of being chosen.

But what does the probabilities returned represent if the dependent variable is Ranks, frequencies or proportions?

For example I have this model:

NLOGIT ; Lhs = BFRANK, CT
    ; Rhs = v
    ; Prob = probs 
    ; Ranks $

Where 1 = "best",2,3....etc

Is probs the probability of being best?

Similarly for frequencies, is probs the probability of highest frequency, and for proportions, the biggest proportion?

If not I can't work out what they signify.

Thanks for any input.