Hannah,
the test tells you that you're odds depend of other outcomes and, therefore, you
are violating the IIA assumption. As Scott and Freese (2001) note in their book,
results of the various IIA tests are often inconsistent. McFadden (1973)
suggests that models estimated with -mlogit- and -clogit- should only be used if
outcome categories can be plausibly differentiated and weighed independently by
decision-makers. I also observed with some of our data sets that IIA holds
according to the Hausman test, but is violated in the Small & Hsiao test. For
further insights on that I would be grateful.
Greetings,
Daniel
---------- Forwarded Message ----------
Date: 15 November 2002 15:16 +0000
From: Hannah Kewsberry <[email protected]>
To: [email protected]
Subject: mlogtest
Hi there,
Sorry to send this e-mail to your personal a/c. I have posted a question to
the Statalist this morning, but it didn't appear even 12 hours later. Since
this issue is kind of urgent, I am sending my question to a few people who
ever replied to my inquiries before, hope you don't mind. My question is: I
tried to run Small-Hsiao test for my multinomial logit but the response
says 'against H0', could you kindly suggest where have I possibly have gone
wrong?
**** Small-Hsiao tests of IIA assumption
Ho: Odds(Outcome-J vs Outcome-K) are independent of other alternatives.
Omitted | lnL(full) lnL(omit) chi2 df P>chi2 evidence
---------+---------------------------------------------------------
1 | -3931.826 -3839.075 185.502 15 0.000 against Ho
2 | -3659.149 -3589.573 139.151 15 0.000 against Ho
3 | -3122.853 -3027.661 190.384 15 0.000 against Ho
4 | -3495.731 -3432.914 125.632 15 0.000 against Ho
5 | -3232.598 -3135.791 193.614 15 0.000 against Ho
8 | -3350.732 -3261.296 178.871 15 0.000 against Ho
9 | -3325.947 -3228.943 194.009 15 0.000 against Ho
-------------------------------------------------------------------
Thanks in advance, Hannah
_________________________________________________________________
The new MSN 8: advanced junk mail protection and 2 months FREE*
http://join.msn.com/?page=features/junkmail
---------- End Forwarded Message ----------
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
