Timeline for Multinomial Logistic Regression: IIA violated - What to do?
Current License: CC BY-SA 3.0
9 events
| when toggle format | what | by | license | comment | |
|---|---|---|---|---|---|
| Dec 18, 2015 at 0:50 | answer | added | Ryan | timeline score: 1 | |
| Aug 5, 2014 at 14:38 | history | edited | Andy | CC BY-SA 3.0 |
fixed typo, added tag
|
| Sep 23, 2013 at 1:16 | history | edited | Vincent | CC BY-SA 3.0 |
Reply to answers and comments
|
| Sep 18, 2013 at 18:14 | answer | added | dimitriy | timeline score: 4 | |
| Sep 18, 2013 at 14:01 | history | edited | Vincent | CC BY-SA 3.0 |
Replied to comments
|
| Sep 18, 2013 at 7:03 | comment | added | dimitriy | On page 245 of Long & Freese (2006) they write that negative test statistics are common. They also cite Hausman & McFadden (1984, p.1226) who say a negative value means IIA was not violated. The absence of p-values is puzzling. What does Small-Hsiao IIA test produce? | |
| Sep 16, 2013 at 3:07 | history | tweeted | twitter.com/#!/StackStats/status/379441292713144320 | ||
| Sep 16, 2013 at 2:25 | comment | added | gung - Reinstate Monica |
I don't know Stata, but it seems that the test of Categ_1 is non-significant, and the other two, having negative values for the chi2, must not "meet asymptotic assumptions". Thus, I'm not sure that you can validly conclude that "the data do not meet the IIA assumption". Possibly nothing can be concluded relative to the IIA assumption. Instead, I think you want to find out what the asymptotic assumptions of the Hausman test are, and what it is about your data that violate them.
|
|
| Sep 16, 2013 at 1:27 | history | asked | Vincent | CC BY-SA 3.0 |