Somalia Student Retention Model

<p>Summary<br />
This is the initial public release of the Student Retention Analysis: Multi-Level Logistic Regression dataset and code package supporting the external evaluation of the Bar ama Baro (BAB) accelerated basic education (ABE) program in Somalia. The release publishes a de-identified analytic subset of the broader longitudinal study, the primary R analysis script, and accompanying documentation to enable transparency, replication, and reuse. It underpins a multi-level logistic regression modeling analysis of student retention that accounts for both individual characteristics (e.g., demographics, household literacy) and site conditions (e.g., school resources, road conditions). Learning from this work is intended to inform evidence-based decision-making in Somalia and similar contexts.</p>

<p>This dataset publication accompanies the external evaluation of Bar ama Baro (BAB), an accelerated basic education (ABE) program in Somalia. It synthesizes findings from a two-year longitudinal study examining (1) BAB effectiveness, (2) learning outcomes for diverse students in BAB Levels 1–2 and in formal primary grades 1–3, and (3) lessons learned for sustainability and scale. Mixed-methods data sources include focus group discussions, key informant interviews, student surveys and learning assessments, teacher surveys, and exit surveys with parents/caregivers of non-retained students and non-retained teachers.</p>

<p>What’s included</p>

<p>Data: SomalisStudentRetentionLocation_Dataset.csv — student-level records merged with location-level observations (de-identified analytic subset).<br />
Code: SomaliaStudentRetentionModel_Code.R — reproducible workflow for bivariate screening and multi-level logistic models (glmer).<br />
Documentation: SomaliaStudentRetention_README.md (project overview, usage), SomaliaStudentRetention_Metadata.xlsx (variable definitions/data dictionary).</p>

<p>Study coverage</p>

<p>Baseline fieldwork: Oct – Nov 2021<br />
Endline (Year 2) fieldwork: Apr – May 2023<br />
Geography: Somalia<br />
Primary outcome: Student retention (binary)</p>

<p>Intended use<br />
Designed for researchers, evaluators, and policymakers interested in early-grade/ABE retention dynamics, program effectiveness, and contextual factors influencing persistence. Suitable for replication, sensitivity analyses, and methodological extension.</p>

<p>Planned updates</p>

<p>Link to baseline, midline, and final datasets (with DOIs) on PURR.<br />
Publication of the full study dataset with expanded variables (subject to disclosure review).</p>

<p><strong>Funding Acknowledgement</strong></p>

<p>This dataset was made possible through support by  USAID/Somalia through USAID/IPI/ITR/R’s Long-Term Assistance and Services for Research (LASER) PULSE Program under the terms of Cooperative Agreement No. 7200AA18CA00009. This activity is undertaken by a multi-disciplinary research team through the LASER PULSE consortium members, with Purdue University leading the effort, in collaboration with Makerere University and Somali Research and Development Institute (SORDI) in Somalia, a member of the Makerere University’s Resilient Africa Network (RAN). The authors also acknowledge the invaluable contributions of the implementing partner, Creative Associates, International, the Somalia Ministry of Education, Culture, and Higher Education, and the funder, USAID/Somalia, to this work.   </p>