Stijn Kelchtermans and Frank Verboven, "Participation and Study 
Decisions in Higher Education", Journal of Applied Econometrics, 
Vol. 25, No. 3, 2010, pp. 355-391.

The data used for the paper come from the Flemish Ministry of Education.
These data were enriched with travel distances and travel times for students
from their home address to each higher education campus (calculated using
Microsoft Mappoint). The data contain the higher education choices of the
cohort of secondary school graduates in 2001.

The paper uses the following data files:

- 54alt_rs14464pup_rsallaltf.zip: 120 variables and 781,056 observations.
The file contains a random sample of pupils and the complete choice set for
each pupil. Used in the aggregate models.

- 155alt_rs12690stud_rs19altf.zip: 236 variables and 253,800 observations.
The file contains a random sample of 12690 students with 20 sampled higher
education study options for each student (the actual choice is always part
of the choice set). Used in the disaggregate nested logit model.

- 155alt_rs19237pup_rsallaltf.zip: 178 variables and 2,981,735 observations.
The file contains a random sample of pupils and the complete choice set for
each pupil. Used in the disaggregate nested logit model.

Each zipfile contains a comma-separated-values datafile with the same name,
in DOS format. Each file is set up in conditional logit format i.e. there
are as many rows for each individual as there are options in the choice set.

The variables in the datafiles are the following:
- INDIV: ID number for the pupil
- OPTION: ID number for the higher education option
- WEIGHT: weight used by the government to determine the variable subsidy
          for the study option
- LSIZE: log of number of study programs offered within the choice alternative
- ACAD/VOCL/VOCS/NONE: higher education type
- SCI/SSCI/BIOM/ARTS: higher education study field
- CATH: catholic higher education institution
- I_*: institution dummies
- CTRAV & CTR61: travel costs below and above the cut-off point for
                 commuter versus resident
- MALE: gender dummy
- FORE: dummy for foreign nationality (non-Belgian)
- AGE: number of years of delay in starting higher education
- ASO: dummy for general secondary education
- TSO: dummy for technical secondary education
- KBSO: dummy for arts/professional secondary education
- CATH1: catholic secondary school
- CLAS: classical language focus in general secondary school 
- LANG1: modern language focus in general secondary school 
- ECON1: economics focus in general secondary school 
- SCIE: science focus in general secondary school 
- MATH: math focus in general secondary school 
- TSOH: product focus in technical secondary school

Students' personal characteristics (MALE etc.) are interacted with higher
education choice characteristics (ACAD etc.). The naming convention for
these interactions is to concatenate the name of the higher education
characteristic with the name of the student characteristic. E.g. ACADMALE
equals 1 for rows that correspond to male students and academic study
options. The interactions of personal characteristics with travel costs
(CTRAV) follow the same logic.

The file estmodelv75c.zip is a Gauss 6.0 computer program. Please refer to
the comments within the program for guidelines on how to use it to estimate
the three models in the paper, as well as how to run the simulations. The
program is an ASCII file in DOS format.