Reproducibility package for How To Build A Reader: Evidence From A Scalable Literacy Intervention In Ghana
Overview
Addressing the massive test score gaps between rich and poor countries will require programs that are both high-impact and scalable. We use an RCT in low-fee private schools in Ghana to study a program that meets both needs. Tools for Foundational Learning Improvement (TFLI) increases test scores by 0.5 SDs after just 9 months of intervention. We use a machine-learning method to decompose the effects by students’ predicted test scores if they did not receive the treatment, and show that the gains are larger for weaker students. Moreover, we show that TFLI’s impacts scale roughly linearly with time as compared to a shorter-term, smaller-scale pilot RCT. The program’s developers use generative AI to accelerate lesson plan development and adaptation to new settings. An observational pilot test of this adaptation to Uganda yields comparable results to our RCT. We develop a model in which basic skills constrain the development of advanced skills, which predicts the pattern of effects we observe across early reading capabilities, and makes forecasts about the future impacts of the program as it continues into second grade
Reproducibility Package
| Repository name | URI |
|---|---|
| Reproducible Research Repository (World Bank) | https://reproducibility.worldbank.org |
Reproducibility
Paper exhibits were reproduced on a computer with the following specifications:
• OS: Windows 11 Enterprise
• Processor: Intel(R) Xeon(R) Gold 6226R CPU @ 2.90GHz
• Memory available: 32.0 GB
Runtime: 12 hours
To reproduce the findings in this package, please follow the steps below:
- Download the data: Once the embargo is lifted, download the required datasets. Refer to the README for detailed instructions on how to obtain the data and the required folder structure.
- Run the code: Open
run_all.RinReplication package.Rproj, update therootpath to your local directory and add paths to your Stata 18 binary and Stata 19 binary and run the entire script. This will generate all the outputs inresultsfolder. There is no need to open Stata separately; the script will automatically launch Stata and execute the required Stata do-files.
In the meantime, the outputs from the replicator’s run of the code are included in this package so users can review the figures and compare them with those published in the paper.
Data
All data is temporarily embargoed by the authors (expected to be made public in the future).
Description
| Author | Affiliation | |
|---|---|---|
| Jason Kerwin | University of Washington | jkerwin@uw.edu |
| Erik Andersen | University of Washington | eander46@uw.edu |
| Simon Graffy | Inspiring Teachers | simon.graffy@inspringteachers.org |
| Monica Lambon-Quayefio | University of Ghana | mplambon-quayefio@ug.edu.gh |
2026-07-02
Scope and coverage
| Location | Code |
|---|---|
| Ghana | GHA |
Disclaimer
The materials in the reproducibility packages are distributed as they were prepared by the staff of the International Bank for Reconstruction and Development/The World Bank. The findings, interpretations, and conclusions expressed in this event do not necessarily reflect the views of the World Bank, the Executive Directors of the World Bank, or the governments they represent. The World Bank does not guarantee the accuracy of the materials included in the reproducibility package.
Access and rights
| Name | URI |
|---|---|
| MIT License | https://opensource.org/license/mit |
| World Bank IGO Rider | https://github.com/worldbank/metadata-editor/blob/main/WB-IGO-RIDER.md |
Contacts
| Name | Affiliation | |
|---|---|---|
| Jason Kerwin | University of Washington | jkerwin@uw.edu |
| Reproducibility WBG | World Bank | reproducibility@worldbank.org |
Information on metadata
| Name | Abbreviation | Affiliation | Role |
|---|---|---|---|
| Reproducibility WBG | DECDI | World Bank - Development Impact Department | Verification and preparation of metadata |
2026-07-02
1