Inequality of Educational Opportunities -- remedial classes in Tanzania “Uguaglianza delle Opportunità Educative: Interventi di Recupero Educativo in Tanzania”

This program is designed to enhance the academic performance of vulnerable primary school students in Tanzania by implementing targeted remedial classes. Focused on students in grades 3 to 7, the intervention provides additional support in key subjects such as Language (Kiswahili), Mathematics, and English to those identified as at risk of falling behind. Through small-group sessions held three times a week over six months, students receive personalized attention aimed at strengthening foundational knowledge and addressing specific learning gaps. The intervention not only focuses on improving immediate academic outcomes, such as test scores and subject mastery, but also emphasizes boosting students' confidence, engagement, and long-term commitment to education. By fostering improved academic performance, reducing dropout rates, and increasing student retention, this initiative aims to create more equitable educational opportunities for marginalized children, paving the way for their future success and socio-economic inclusion.

2024-12-15

2025-07-31

Outcomes of interest include improvements in test scores, attendance rates, and long-term retention in school.

This randomized controlled trial evaluates the impact of remedial classes on improving educational outcomes for vulnerable primary school students in Tanzania. Targeting grades 3 to 7, the intervention provides additional academic support in core subjects, including Mathematics, Kiswahili, and English, for students identified as academically at-risk. Eligible students are randomly assigned to treatment (remedial classes) or control (standard classroom instruction) groups. The intervention involves three 40-minute sessions per week for six months, focusing on reinforcing foundational knowledge in smaller groups. Key outcomes include academic performance, school attendance, and retention rates, contributing to more equitable educational opportunities.

Not available

The randomization process for this study was carefully designed to address both logistical constraints and the ethical considerations of working with vulnerable populations. The sample comprises approximately 5,000 students identified as vulnerable based on their academic performance and socio-economic conditions, as defined by the WeWorld Foundation’s Protocol for Identifying Children at Risk of Dropping Out. However, due to limited resources, only 2,500 of these students were enrolled in the remedial classes. An additional 1,500 students, despite meeting the vulnerability criteria, could not be included in the intervention due to capacity limitations.
To ensure fairness and minimize selection bias, the 2,500 students who participated in the intervention were randomly selected from the pool of identified vulnerable students. This process was conducted within each classroom and grade level to maintain representativeness and balance across different subgroups. The selection process was conducted randomly within each school to ensure that the participants' names were drawn to ensure equal opportunity to participate in the remedial program.
The control group was carefully constructed to include two distinct sets of students for robust comparative analysis. First, a control group was drawn directly from within the same classes as the treatment group, ensuring that both groups share similar classroom dynamics, teacher exposure, and school environments. This intra-class approach allows for a precise comparison by minimizing potential external confounding factors. Second, an additional group was formed from the larger pool of approximately 5,000 vulnerable students who met the criteria for remedial education but could not be enrolled in the intervention due to resource constraints. These students represent a critical segment that highlights the demand for remedial support. Including them as a secondary control group provides insights into the broader impact of limited access to the intervention, offering a more comprehensive evaluation of the program's potential scalability and equity implications.

The randomization in this study is conducted exclusively at the individual level, focusing on eligible students identified as vulnerable within the same classrooms across ten participating schools. Students are selected based on predefined academic criteria, ensuring they meet the vulnerability threshold for participation. From this pool, individuals are randomly assigned to either the treatment group (remedial classes) or the control group (no remedial classes but regular classroom instruction). This individual-level randomization ensures that within each classroom, all students, regardless of their assigned group, share the same educational environment, including teachers, curriculum, and general school resources. This approach allows for a direct comparison of outcomes between treated and untreated students, isolating the impact of the remedial classes while controlling for external classroom factors. By restricting randomization to individual students rather than grouping by schools, the study achieves a higher degree of precision in measuring the intervention’s effects. Moreover, this method captures the diversity of classroom dynamics across the ten schools involved, ensuring that findings reflect the heterogeneity of the broader educational context. This strategy strengthens the internal validity of the study while providing robust and generalizable insights into the impact of remedial education interventions.

Yes

The intervention is structured around approximately 40 classes across 10 schools, with each class serving as an independent cluster. Clustering at the class level allows for a detailed analysis of how the intervention impacts students within a more localized and controlled educational environment. This approach ensures that variations stemming from teacher practices, classroom resources, peer interactions, and other within-class dynamics are appropriately captured and controlled for in the analysis. This level of granularity is crucial for isolating the intervention's direct impact on students' academic performance and other educational outcomes, minimizing the potential for school-level confounding factors.
Additionally, this class-level clustering allows for a more equitable distribution of the intervention across the 10 participating schools. It ensures that each school's unique characteristics -- like infrastructure, community engagement, and overall academic culture -- do not unduly influence the outcomes pointing out aspects related to the internal and external validity of the study, providing insights that are not only robust but also generalizable to similar educational settings.

The study plans to include approximately 5,000 students, selected based on their vulnerability status. From this pool, 2,500 students will participate in the intervention (treatment group), while the remaining will serve as the control group in order to promote a robust comparison of outcomes across treatment arms.

The study includes 10 schools, encompassing a total student population of approximately 20,000 children. From this larger group, a subset of around 2,500 students, identified as highly vulnerable based on the WeWorld vulnerability index, will be selected for participation. These 2,500 students are randomly drawn from a broader pool of 5,000 vulnerable children identified across the schools. The exact number of participants chosen from each school will depend on the distribution of vulnerability scores, ensuring that the intervention is focused on those most in need while maintaining proportional representation across schools.
The treatment group will consist of students receiving remedial classes, while the control group will include students from the same classrooms who share similar levels of vulnerability but will not receive the intervention. This design ensures that the control group is drawn from the same educational and social environment as the treatment group, maintaining comparability. To be more specific, the intention is to include all classes with students participating in the treatment group. While the exact size of these groups may vary slightly depending on the number of eligible students identified within each school, the randomization process ensures a balanced and equitable representation of treatment and control groups across all participating schools.

The study aims to detect a minimum effect size of 0.15 standard deviations in key academic outcomes, such as exam scores, attendance rates, and retention rates. This calculation assumes a confidence level of 95% and a power of 80%, accounting for the clustering within schools and the stratification by classroom and vulnerability status. The sample size and design are optimized to ensure the detection of meaningful improvements attributable to the remedial classes, providing actionable insights for future policy and programmatic decisions.