Is mlogit appropriate? Hausman test of IIA results and interpretation

[Jill Adona]

Dear all,

The subject of post-high school choices has been studied extensively using the multinomial logit model. I'm estimating it for my thesis, but I'm concerned about the results of the Hausman test of IIA that I have been getting.

1. Note: the rank of the differenced variance matrix (51) does not equal the number of coefficients being tested (56); be sure this is what you expect, or there may be problems computing the
test. Examine the output of your estimators for anything unexpected and possibly consider scaling your variables so that the coefficients are on a similar scale.

2. No standard errors

3. chi2(51) = (b-B)'[(V_b-V_B)^(-1)](b-B)
= -20.02 chi2<0 ==> model fitted on these
data fails to meet the asymptotic
assumptions of the Hausman test;

Can anyone suggest a solution to the problems? I'm also concerned about the interpretation of the chi-squared, but perhaps it is because of problems 1 and 2. If the IIA isn't violated in a lot of my related literature, is it possible that it is with my data (survey data, with 5 outcomes)?

Really hoping to receive some help. Thank you!

[Jill Adona]

I solved it. I overlooked that I used pw in the unrestricted model then I was using iw in the restricted. Sorry, everyone!

Although Problem 1 appears in a different time period. Any suggestions?

Dear all,


1. Note: the rank of the differenced variance matrix (51) does not equal the number of coefficients being tested (56); be sure this is what you expect, or there may be problems computing the
test. Examine the output of your estimators for anything unexpected and possibly consider scaling your variables so that the coefficients are on a similar scale.

Really hoping to receive some help. Thank you!