st: Why to use and how to interpret interaction terms in nlogit?

[Patricia Yu <pyu1@wisc.edu>]

Dear Statalist,

 

I learn the nested logit model by running –nlogit- with a travel dataset which could be obtained by typing the following in Stata:

 

. webuse travel

 

Tree structure for the nested logit model is below:

 

top ---------> bottom

 

 

In this data exercise example, 4 dummies describing choices/modes, respectively, are created: named <air>, <train>, <bus>, and <car>.

An interaction term of “choosing air” with “household income” is also created: named <air_inc>.

The result table of nlogit is presented as follows:

 

 

My questions are:

1.    Why 3 dummies, <air>, <train>, and <bus>, which are alternatives of choice, are put in the model as independent variables? How to explain the coefficients of these three dummies?

2.    Why the interaction term of “choosing air” with “household income” is created for the second equation (the equation of type)? How to explain its result?

 

Thank you very much in advance for helping me clarify these!

 

Thanks,

Patricia

 

-------------------------------------------------------------------------

PATRICIA YU

Dissertator in Higher Education

Department of Educational Leadership & Policy Analysis

School of Education

University of Wisconsin-Madison

(608) 698-8108

pyu1@wisc.edu

-------------------------------------------------------------------------