I am using asclogit to analyze a choice model (12 cases with 3 alternatives, 1 independent continuous variable price, 2 independent binary variables), and the coefficients I receive are negative. While I would expect the price to be negative, the two other binary variables should be positive according to my expectations. So when I estimate a willingness to pay for the binary variables it thus returns a negative value.

Wondering if the note I received at the beginning of running the command could be the reason for this fault: variable binary-1 has 1795 cases that are not alternative-specific: there is no within-case variability

If that's not it, what other reasons could cause my model to return negative instead of positive coefficients?

Thanks!