Sander Gerritsen, Erik Plug, and Dinand Webbink, "Teacher Quality and
Student Achievement: Evidence from a Sample of Dutch Twins", Journal of
Applied Econometrics, Vol. 32, No. 3, 2017, pp. 643-660.

The workfile (1842 observations, 921 twinpairs) has been constructed
from 2 files: 

1) PRIMA-files (PRIMA-cohorten Basisonderwijs 1995-2005), open access:
http://www.dans.knaw.nl/en/search

2) Twin-files in PRIMA. We asked the constructers of PRIMA to give us
the twincodes: those with same name and date of birth.  These files
were given confidentially but are not privacy sensitive. They are
available upon request by sending an email to sbg@cpb.nl or to the
builders/constructers of PRIMA (Kohnstamm institute):
secr@kohnstamm.uva.nl

By matching these two files one could easily destill our workfile:
twins that go to the same school but are assigned to different class
rooms.  Class rooms in the Netherlands are often given names like 2A,
2B (for different classrooms in grade 2)

Codes in PRIMA, our workfile and dofile are all in Dutch. PRIMA
consists of a lot variables. 

I will only translate those variables in English that are used in our
paper, and hence our dofile.  Only these variables are necessary to
replicate the results outlined in the paper.

*taal=test score reading
*reken=test score math
*taal_vor=previous test score reading
*reken_vor=previous test score math
*vrouw_doc=teacher is female
*meerdere_grdoc_doc=class room is taught by two (parttime) teachers
*combi_klas_doc=class room that mixes grades
*t_klas_doc=class size
*Ep_meiden_adm=% girls
*Ep_ned_adm=% natives
*jaren_werkzaam_doc=experience of teacher in years
*groepnr=grade
*schoolnr=school
*groepsna=class room code

Important in the dofile: the D stands for Difference between twins,
hence twin j- twin j'

The file gpw-text.zip contains the workfile in CSV format, and a Stata
file that created it.

The file gpw-text.dta contains the workfile in Stata format.