Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down on April 23, and its replacement, statalist.org is already up and running.


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

st: Hausman IIA test after asclogit in Stata11


From   Simon Falck <simon.falck@abe.kth.se>
To   "statalist@hsphsun2.harvard.edu" <statalist@hsphsun2.harvard.edu>
Subject   st: Hausman IIA test after asclogit in Stata11
Date   Thu, 21 Mar 2013 13:52:29 +0000

Hi Statalist,

I have a question that is not new to Statalist regarding the Hausman IIA test after obtaining results from an alternative specific choice model with 72 choices, implemented with -asclogit-. 

The dependent variable take the value 1 when an alternative is chosen, otherwise 0. I have 10 independent variables. The asclogit model runs well and is specified as:

asclogit Y $xlist, case(id) casevars() alternatives(alt) basealternative() noconstant
est sto full
asclogit Y $xlist if alt!=72, case(id) casevars() alternatives(alt) basealternative() noconstant
est sto partial

Thus, the full model includes all 72 alternatives (full), and the reduced model (partial) excludes one of these alternatives. I get the following results from the hausman test regarding IIA test.

hausman partial full

Test:  Ho:  difference in coefficients not systematic
chi2(13) = (b-B)'[(V_b-V_B)^(-1)](b-B)
                 = 1.59
Prob>chi2 = 0.9999
(V_b-V_B is not positive definite)

It seems there is a problem as the test reports "V_b-V_B is not positive definite". I have looked through previous conversations regarding similar cases but could not find any satisfying answer.

Any suggestions what may be wrong in my specific case?

Thanks in advance
Simon


*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/


© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   Site index