Bookmark and Share

Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at

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

Re: st: Mlogit: Log pseudolikelihood bug?

From   Maarten buis <>
Subject   Re: st: Mlogit: Log pseudolikelihood bug?
Date   Sun, 27 Feb 2011 12:48:22 +0000 (GMT)

--- On Sun, 27/2/11, Robert L. Hellpap wrote:
> I tried to estimate a quiete complex model with more than
> 100 dummy's as independent on a 6 category dependent
> variable. When my results where displayed the log
> pseudolikelihood value was announced as ".z_" which should
> be a missing value. What does ist mean that the log
> pseudolikelihood is a missing value and is it legitime to
> reduce the number of iterations until I receive a legal
> value?

It means that you need to simplify your model. Remember that
the trick of adding dummies to get a fixed effects regression
does not work for non-linear models like -mlogit-, so the
value of adding that many dummies is often doubtful anyhow.

Hope this helps,

Maarten L. Buis
Institut fuer Soziologie
Universitaet Tuebingen
Wilhelmstrasse 36
72074 Tuebingen


*   For searches and help try:

© Copyright 1996–2018 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   Site index