# st: Nested logit model estimation with Stata

 From "Gaston Gohou" To statalist@hsphsun2.harvard.edu, majordomo@hsphsun2.harvard.edu Subject st: Nested logit model estimation with Stata Date Tue, 23 Nov 2004 06:03:50 -0500

```Dear all ,

I would like to run a Nested logit Model .

The data are such as for each observation, I have one possible choice  (the
number of alternative for the choice is 4 for level 1 - bottom level- and 2
for level 2 - top level)  and the dependant variables are related to the
observations, not the alternatives. Here is an example of the row data

id           Choice _top        choice_bottom       age       education  ...

1                 1                        1                      45
3

2                 1                        2                     56
6

3                 2                        1                     39
4....

The top choice has 2 alternatives and each alternative of the top  choice
has 2 alternatives. At the end, I have 4 alternatives (at the bottom).

The dependant variables are linked to the observations, not the
alternatives.

I transformed the data before the procedure nlogit estimation as stated the
the stata manual:

id           Choice _top        decision age_1  edu_1  edu_2  edu_3
choice ...

1                    1                        1     45         3          0
0              1

1                    1                        0     45          0         3
0               2

1                    0                        0     0            0         0
0               3

1                    0                        0     0            0         0
0               4

2                    1                        0     56          6         0
0               1

2                    1                        1     56          0         6
0               2

2                    0                        0     0            0         0
6               3

2                    0                        0    0            0         0
0               4

.....

Below is the code I used:

nlogit chosen  (decision = edu_1 edu_2 edu_3) (choice_top= age_1), group(id)
robust

edu_1~edu_3 are the bottom level dependent  variables and age_1 is the top
level dependant variable

Unfortunately, when I ran this model it doesn't converge and I tries several
specification for the variables. But after a few hours, there is no converge
for the ML procedure.

I have 2 questions:

1. How to set up the data in my case to run this model? Is the way I have
that is correct?
2. How to interpret the result I will get? I know how to interpretre the
data for the dummy regressor I created but what about the other other
variables?

Gaston Gohou
PhD Candidate
University of Maryland, College Park

*
*   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/
```