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

st: RE: AW: MLOGIT versus a set of LOGIT models

From   "Nick Cox" <>
To   <>
Subject   st: RE: AW: MLOGIT versus a set of LOGIT models
Date   Tue, 18 Aug 2009 17:59:14 +0100

Jon's use of -xi:- here is really a side issue. While it's quite true
that factor variables in Stata 11 offer more power and flexibility than
did -xi:- it is equally true that -xi:- remains available and continues
to produce good results. It seems to me that Jon's question remains
essentially untouched by Martin's comment. 


Martin Weiss

In the absence of a -version- statement, your code is problematic as it
still uses -xi- which has been superseded in Stata 11. Note the FAQ on

I think you will get more/better answers if you frame your question in
of the example in the help file... 

Jon Heron

 I have always believed that the regression estimates from a multinomial
 regression model (MLOGIT) could be replicated through a set of simple
 logit models with the appropriately derived binary outcomes.

 Whilst attempting to demonstrate this fact for some teaching material
 polytomous IRT that i am writing, I moved from my usual categorical
 predictors to a continuous covariate + discovered that the above
 did not hold.

 To ensure that this was not a fluke, I have since demonstrated that
 the two simple models below, each with the same 4-level outcome and
 predictor, the former can be replicated with logits, but not the

 xi: mlogit ghq1 i.ghq3, baseoutcome(0)
 mlogit ghq1 ghq3, baseoutcome(0)

*   For searches and help try:

© Copyright 1996–2015 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index