Reproducible code archive accompanying the manuscript "The limited impact of extending the school week on educational outcomes in Dutch primary education: a quasi-experimental study using penalized synthetic control analysis".
This code repository is designed to run on the microdata environment of Statistics Netherlands (CBS). For more information about this, see the microdata webpage.
The project uses R and RStudio. To run the project, install both and then open the file nprz_synthetic_controls.Rproj. Install the dependencies as follows:
deps <- c("haven", "tidyverse", "mice", "purrr", "furrr", "pensynth", "Synth", "patchwork", "writexl")
install.packages(deps)The preprocessing scripts can be run all at once by running the file 01_run_preprocessing.R from the main directory.
RScript 01_run_preprocessing.R
In RStudio, you can use a "Job" for this.
This should take about 1 hour 45 minutes.
Each of the preprocessing steps in the folder preprocessing_scripts can also be run independently.
The scripts 02-06 run and inspect the synthetic control method, as follows:
| Script | Goal |
|---|---|
02_quality_control.R |
Perform several checks on the pre-processed data |
03_summarization.R |
Compute school-level variables from student-level data |
04(a/b)_dataprep_synthetic_control_(cito/isled).R |
Data preparation, missing value handling, preparation of matrices for performing synthetic control |
05(a/b)_compute_synthetic_control_(cito/isled).R |
Compute the synthetic control for each school |
06(a/b)_inspect_synthetic_control_(cito/isled).R |
Inspect the results of the synthetic control procedure, produce plots, and more |
07_prepare_output.R |
Preparation of results for output check at Statistics Netherlands |
In addition, the 04 and 05 scripts have versions in the folder robustness_checks to re-run the analyses with more potential covariates.
This is a project by the ODISSEI Social Data Science (SoDa) team. Do you have questions, suggestions, or remarks on the technical implementation? Create an issue in the issue tracker or feel free to contact Erik-Jan van Kesteren or Gijs Custers.
