School Location, Digital Conditions, and Student Achievement: Evidence from PISA 2022 in Spain
Description
Code, aggregate results and documentation supporting the manuscript “School Location, Digital Conditions, and Student Achievement: Evidence from PISA 2022 in Spain”, revised on 1 October 2026. The study examines associations between student-reported ICT availability, subject-related ICT use and perceived ICT quality and mathematics, science and reading achievement in Spain, comparing schools in communities with fewer than 3,000 residents with other locations. Analyses use all ten plausible values per domain, final student and 80 BRR/Fay replicate weights, and model-specific complete-case samples. The ZIP contains the Stata 19 workflow, aggregate CSV/DTA outputs, Figure 1, variable dictionaries, supplementary diagnostic scripts, execution logs, a table-to-output crosswalk, validation notes and SHA-256 checksums. Numerical results correspond to the revised manuscript. The package was validated on 1 October 2026; the retained full estimation run was completed on 27 July 2026. PISA student-level and school-level microdata are not redistributed. Users must obtain source data through the OECD under its access terms. Re-estimation requires the separately obtained, correctly prepared analytical input documented in the README (30,800 students and 153 variables). The original student-school merge script and exact upstream release are unavailable, so this archive is not a self-contained reconstruction from raw OECD files. Legacy code labels RURAL and CCAA are retained for compatibility; CCAA covers 17 autonomous communities plus Ceuta and Melilla. All reported associations are observational and should not be interpreted as causal effects. This version supersedes the earlier analytical specification in Version 1. See README_Mendeley.md, SOURCE_DATA_NOTICE.md and VERSION_NOTES.txt for scope and changes.
Files
Steps to reproduce
1. Download and extract ICT_PISA2022_Spain_Code_and_Results.zip. Read README_Mendeley.md and SOURCE_DATA_NOTICE.md. 2. To verify the supplied files and aggregate calculations without Stata or microdata, run: python3 verify_package.py. The archive includes aggregate results and variable metadata, not individual-level PISA records. 3. For re-estimation, obtain PISA 2022 source data through https://www.oecd.org/en/data/datasets/pisa-2022-database.html and comply with the linked PUF access terms. Independently prepare the required Spain student-school input using the supplied codebook, value-label dictionary and provenance script. The original upstream merge and exact source release are unavailable; successful raw-file reconstruction has not been verified. Create_Mendeley_Replication_Data.do requires the original merged project file and is not an automatic OECD downloader or merger. 4. Place the correctly prepared Mendeley_Replication_Data.dta beside Revised_Analysis.do. The README documents the reference input SHA-256, 30,800 observations and 153 variables. Use a working copy because the do-file regenerates analysis_output/. 5. In Stata 19, install repest and pv using ssc install repest and ssc install pv; change directory to the extracted folder and run do Revised_Analysis.do. Audited dependency versions and hashes are listed in the README. The historical full run took about 67 minutes; runtime depends on the machine. 6. Optional Python metadata/diagnostic regeneration requires pandas==2.2.3 (requirements.txt). With the external input in place, run python3 build_value_labels.py and python3 build_supplementary_diagnostics.py. The October check reran data preparation and deterministic diagnostics and reconciled manuscript values with stored outputs; it did not rerun all statistical models. Appendix A.4 reports within/between interaction contrasts, not every simple slope.