[Limdep Nlogit List] questions about nested logit model

[xuan]

Dear all,

I am learning LIMDEP to using discrete choice model technique to solve
problems in transportation. I have several questions about dataset design.

Recently, I read a journal paper at this link:
http://www.bts.gov/publications/journal_of_transportation_and_statistics/volume_05_number_23/html/paper_02/

Which presented three alternative structures. They all have a common
characteristics: more than one alternatives in the lowest level of the
nested logit model can be shared by one individual.

For instance, the second model structure in the paper:

[image: cid:image001.png at 01CAD01B.F6855CB0]

The data for model estimating is the 48-hour trip information responsed to
the 1996 San Francisco Bay Area activity survey. I found that one
individual’s information can reflect more than one choice, such as “eat at
work” and “entertainment after work”.



My question is how this kind of dataset looks like. I really don’t
understand how to design the dataset for estimating a nested logit model. I
am a beginner of LIMDEP, the dataset for estimating the discrete choice
model is multiple lines representing one individual and always with only one
choice variable of value “1” indicating the actual choice.

Thank you for helping me!

Best!

PS: the figure also can be seen in the attachment.


Xuan