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:A question related to mixlogit package

From   Arne Risa Hole <>
Subject   Re: st:A question related to mixlogit package
Date   Fri, 1 Jun 2012 09:58:43 +0100


The message means what it says; variables must vary over
groups/alternatives to be included in the model. This is also true for
the standard conditional logit which can be estimated with -clogit-
and/or -asclogit-. You should always start with these models first
before trying to run a mixed logit – if you can't get -clogit- to run
you can be 100% sure that -mixlogit- won’t either so this is a good
way to discover fundamental problems with your model spec.


On 1 June 2012 09:11, Nina YIN <> wrote:
> Dear all,
> I am currently using comand( mixlogit) to estimate mixed logit model
> with individual-level data and have some questions to ask for your
> help.
> I want to estimate the demand of prescription drug with data of
> purchasing records, drug characteristics and individual demographics.
> When I use the command,
> -mixlogit choice price, group(ncs) id(np) rand(drug characteristics,
> individual characteristics, interaction terms)
> The results always return with" Variable education has no within-group
> variance" or "Variable age has no within-group variance".
> In my sample, there have some individual has repeated purchasing,
> therefore, as a start, I excluded all the repeated records such that
> for each individual there is only one purchasing record. However, it
> also returned with this message. I can't get what does it mean. Does
> that mean in this type of model, I couldn't include the individual
> attribute variables except the interacted terms?
> If you ever have had such experience, could you share me with your
> opinion? Thanks a lot!
> --
> Best Regards,
> Nina YIN
> Toulouse School of Economics
> Manufacture des Tabacs
> 21 Allee de Brienne
> Toulouse, 31000, France
> Tel: 0033-(0)5 6123 8348
> *
> *   For searches and help try:
> *
> *
> *

*   For searches and help try:

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