Replication code: Large learning gains from government-delivered school reform in Morocco
# Large learning gains from government-delivered school reform in Morocco
This repository contains the Stata and R code that reproduces the tables and
figures for a study of the one-year effects of Morocco's Pioneer School Program
(PSP), a government-implemented reform in public primary schools that combines
structured pedagogy with targeted remediation following the "Teaching at the
Right Level" approach. The evaluation uses a prospective, pre-registered matched
difference-in-differences design; learning outcomes are item-response-theory
(IRT) scored assessments in Arabic, French, and Mathematics; heterogeneity is
estimated with a causal forest.
The entry point `code/00_master.do` reproduces every exhibit from analysis-ready
("clean") data.
## Preregistration
The pre-analysis plan for this study is registered on the Open Science
Framework: .
## Data availability
The datasets that support this study are **not** included in this repository and
are provided separately, subject to their ownership and access conditions.
- **Primary data (author-collected).** The baseline and endline student
assessment data (the item responses underlying the IRT-scored Arabic, French,
and Mathematics tests) were collected by the authors for this study. These can
be shared for peer review and, after a deidentification pass, will be
deposited openly.
- **Administrative data (restricted).** All other datasets (school
administrative records, examination scores, student enrollment and dropout
records, student lists, and the matched school sample derived from them) are
owned by Morocco's Ministry of National Education. They cannot be
redistributed and can be made available only to the journal's Data Editor
under a data-use agreement.
The analysis reads the primary data through the global `$published` and the
administrative data through `$restricted` (both set in `paths.do`). Neither
directory is part of this repository. Raw source data and the upstream
cleaning/deidentification are no …