Replication materials for online victimization and academic achievement in PISA 2025

Description

This dataset provides code, aggregate results and replication instructions for a PISA 2025 study of 420,859 students across 69 education systems. The study examines how the frequency of upsetting information posted online about students without their consent relates to achievement in mathematics, reading and science. Sensitivity analyses and additional online victimization measures are included. The package reproduces results and figures from aggregate inputs. Re-estimating the models requires OECD microdata, which are not included. A complete rerun using the authors’ prepared extracts reproduced all 18 result tables. Instructions for obtaining and preparing the source data are provided, although these initial steps were not retested. The findings describe adjusted associations, not causal effects

Steps to reproduce

Unzip the archive and open a terminal in the Online_Victimization_Replication directory. Use Python 3.12 or 3.13, create a virtual environment, and install the dependencies with: python -m pip install -r requirements.txt Run: python run_replication.py verify Then run: python run_replication.py aggregate --output ../replication_output --figures This reconstructs 14 numerical result tables from the included country-level sufficient statistics and regenerates the 19 statistical figure panels. The output folder must be new. This workflow requires no student or school microdata. To refit all models, follow SOURCE_PREPARATION.md to obtain the specified OECD PISA 2025 files and prepare analysis_input.pkl and ict_items_extraction.pkl. Source preparation requires Stata 17 or later; the statistical analysis runs in Python. Then run: python run_replication.py microdata --output ../microdata_output --analysis-input /path/to/analysis_input.pkl --ict-input /path/to/ict_items_extraction.pkl --figures Newly serialized but exactly value-equivalent extracts can be used with the documented --accept-equivalent-serialization option. The complete microdata refit was verified using the authors' exact local extracts and reproduced 18 numerical tables. Fresh OECD downloading and upstream conversion were not rerun during packaging. The separate graphical abstract is not generated by these commands.

Institutions

• Universidad de Zaragoza

  Aragon, Zaragoza
• University of Patras

  West Greece, Pátrai
• Universidad de Sevilla

  Andalusia, Seville
• Universidad de Costa Rica

  San José, San José

Categories

Funders

• Gobierno de Aragón

  Aragon, Zaragoza

  Grant ID: S44_23R

Related Links

Licence