negative Hausman statistics in logit panel & the robust Hausman test in logit panel

[Mohammad Nasrollahi]

Dear all,
I am performing a Hausman test on logit panel data to determine whether to choose Random Effects or Fixed Effects for my analysis. After performing the test I get this error:

chi2(10) = (b-B)'[(V_b-V_B)^(-1)](b-B)

= -1.67 chi2<0 ==> model fitted on these

data fails to meet the asymptotic

assumptions of the Hausman test;

see suest for a generalized test

what to do in this case? (by negative Hausman statistics in logit panel)

what the robust Hausman test in logit panel?

Thank you very much in advance!

Mohammad

[Mohammad Nasrollahi]

Has nobody any idea how to solve my problem?

I would appreciate any help or hints.

[Pia Schmidt]

I had the same problem as you and found this topic:

https://www.statalist.org/forums/for...-negative-chi2

Maybe this will help you.

[Carlo Lazzaro]

Mohammad:
Pia points you to the only recent thread I know that considers this tricky topic.

[Mohammad Nasrollahi]

Thanks a lot, Carlo and Pia.
The problem is negative Hausman statistics in logit panel and the use of sigmamore, sigmaless & xtoverid options is not possible. Moreover, based on post # 6 (from Carlo), can this negative amount really be interpreted as confirming the zero hypothesis in logit panel? I really need help.

[Carlo Lazzaro]

Mohammad:
this is the best source I've found.
Hence, I would go -xtlogit,re-.

[Mohammad Nasrollahi]

Thank you very much for your response Carlo.