Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down on April 23, and its replacement, is already up and running.

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

st: Gllamm multinominal logit model estimation problems

From   Alec Minnema <>
To   <>
Subject   st: Gllamm multinominal logit model estimation problems
Date   Mon, 13 Feb 2012 12:04:15 +0100

Dear Statalist users, 

I am trying to estimate the choice of households to buy a product i in week t. I would like to do this with the gllamm module in Stata.

The data is in the expanded format and consist of the following variables:

Brand= The set of brands, in total 4 brands.

Brand1-Brand3 are dummy variables for the brands.

v35= Price of brand x in week t.

v37=The variable that identifies consist of the ID(v5) and the week t.

BrandSelected=Indicates if the brand is selected


The resulting model in Gllamm is specified in the following manner:


gllamm Brand Brand1 Brand2 Brand4 v35, expanded(v37 BrandSelected o) i(v5) noconstant iterate(20) nip(20) family(binom) link(mlogit) adapt trace search(20) 


Please note that there are atleast 5 observations per ID, so in 5 weeks the households decides to buy a brand.

Please note that there are no none options.

Please note that a household can only buy one brand per week.

In total there are:

number of level 1 units = 1444
number of level 2 units = 37

If I estimate this model, it first finds initial values in four iterations.

Afterwards, stata produces the following output:

start running on 13 Feb 2012 at 11:51:57

Searching for initial values for random effects
 -453.4502  -453.4502  -453.4502  -453.4502  -453.4502  -453.4502 
 -453.4502  -453.4502  -453.4502  -453.4502  -453.4502  -453.4502 
 -453.4502  -453.4502  -453.4502  -453.4502  -453.4502  -453.4502 
 -453.4502  -453.4502 
Running adaptive quadrature
Iteration 0 of adaptive quadrature:
Initial parameters:

        Brand:     Brand:     Brand:     Brand:       v51:
       Brand1     Brand2     Brand4        v35      _cons
y1  -.4985762  -.7274367    .344051  -.0191824         .5

Updated log likelihood:
 -453.4502  -453.4502 
                                                   log likelihood =  -453.4502
Iteration 1 of adaptive quadrature:
Updated parameters:

(error occurred in ML computation)
(use trace option and check correctness of initial model)
finish running on 13 Feb 2012 at 11:52:02


Can anybody please help me by explaining the reason for this error term and how to solve it.

Kind Regards,

Alec Minnema

PhD student, University of Groningen.

*   For searches and help try:

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