# st: Nested logit estimation

 From Subhash.Pokhrel@urz.uni-heidelberg.de To statalist@hsphsun2.harvard.edu Subject st: Nested logit estimation Date Mon, 17 Nov 2003 21:58:43 +0100

```Dear Listmembers,

I have the following data set (there are many more variables and observations;
this is an example only):

id      time     income    restaurant       kid
1.    10105        15   3931.572                       3
2.    10106        15   3931.572                       3
3.    10107        15   3931.572     Freebirds         3
4.    10206        30     4289.5    MamasPizza         4
5.    10207        30     4289.5                       4
6.    10208        30     4289.5       Madcows         4
7.    10210        30     4289.5                       4
8.    10405        15   6843.571  Christophers         3
9.    10406        15   6843.571     CafeEccel         3
10.    10407        15   6843.571                       3
11.    10504        30     7223.5   Losnortenos         3
12.    10505        30     7223.5                       3
13.    10506        30     7223.5                       3
14.    10710        30   13809.57    Wingsnmore         1
15.    10805        30   2809.111                       4

I want to model the decision to eat at one of the restaurants using nlogit. My
problem differs from the famous STATA example as follows:

Not all individuals in the sample go to a restaurant for dinner. Some may not
feel hungry (i.e. they skip their dinner for this evening), those who feel
hungry may wish to cook at home. Only those who decide to go out choose one of
the six restaurants. The missing values in restaurant column above therefore
represent either "no-hungry" individuals or "love-homefood" individuals.
Typically, the decisions are:

Stage 1: Hungry (0=No; 1=Yes)
Stage 2: Eating place (0=Cook at home; 1=Go to Restaurant)
Stage 3: Restaurant (1= MamasPizza; 2=Freebirds; .........7=MadCows)

Does anyone help me how to generate nlogittree for this problem, and
subsequently estimating it? Do I need to reshape the data to long format
before generating the tree?

Thanks. I would appreciate receiving (preferably a quick) feedback.

Subhash.

*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/
```