image image
  • About us
  • projects
  • Team
    • Core Team
    • affiliates
  • Publications
  • Career
  • Contact us
#

In many developing countries, teachers often misjudge their students' abilities, leading to ineffective teaching and leaving struggling students without the support they need. To address this, the World Bank, in collaboration with the Government of Bangladesh, conducted a study to determine whether diagnostic feedback and teacher training could enhance student learning. The World Bank Group and New York University funded this research, which was carried out by the ARCED Foundation in partnership with the Secondary Education Development Program (SEDP) and Teach for Bangladesh. The study aims to provide evidence-based insights into the role of formative assessments and teacher training in improving student learning

Tags: Teacher Training, Low-Achieving Students

Teachers Knowledge of Students Skill (TKSS)

Students in developing countries vary widely in preparation for schooling. As a result, teachers may not hold accurate beliefs about the academic skills of their students. This is likely to impact the quality of instruction in the classrooms negatively, and hence affect students’ learning levels.

To address the issue of a lack of teachers’ awareness about their students’ skills, the World Bank, with the support of the Government of Bangladesh, was conducting an impact evaluation in a Randomized Control Trial (RCT) aimed at improving teachers’ knowledge and students’ learning levels.
The study aimed to evaluate whether providing teachers with training and materials, including student report cards and instructional guidance, could improve student learning outcomes.

468 schools participated, divided into two treatment groups and one control group (156 schools each). The 312 schools in the treatment arms received these interventions following baseline and midline assessments.

Interventions and Randomization: We randomly assigned 450 schools to

  1. A control group, which did not receive any interventions.
  2. An external feedback group (T1), which received two formative assessments administered by the research team and reports summarizing their results at baseline and midline, along with training for teachers on how to interpret the assessment results.
  3. An internal feedback group (T2), which received two formative assessments administered by teachers at baseline and midline, along with training for teachers on how to administer and interpret the written assessments and on how to conduct one-on-one oral assessments for the three lowest-ranked students in the written assessment
Objective: The key objectives are

  1. Assess whether teachers hold inaccurate beliefs about their students’ academic abilities
  2. Compare the effectiveness of external (T1) vs. internal (T2) feedback mechanisms on student learning outcomes.
  3. Determine whether the interventions lead to measurable improvements in math and reading skills among Grade 6 students.
  4. Provide evidence-based recommendations for integrating formative assessments and teacher training into Bangladesh’s national education strategy.
Data Collection and Research Activities 

Baseline Survey

In May 2022, the research team initiated the baseline data collection. 468 schools were selected from three divisions, each clustered within four districts. From each school, 20 students were randomly chosen for surveys and assessments.

Each student completed a 15-minute digital survey to provide background information. This was followed by all 468 schools participating in a 180-minute group-based paper assessment in math and reading. To process the large volume of data, an in-house digitization and scoring team, consisting of approximately 100 operators, was set up. In addition to this core assessment, an extra 30-minute paper-based written test was administered in the 312 treatment schools (T1 and T2) to capture more detailed student learning metrics.

Teachers also participated in the study, with one math teacher per school completing a 45-minute survey on instructional practices and student performance.

This phase established the foundation for measuring student learning and teacher perceptions before implementing interventions.

Unannounced Visits (TBD)

As the study progressed, the research team conducted unannounced school visits to assess classroom engagement and verify attendance data. These visits occurred between November 2022 and February 2023, covering all 468 schools in the study. On each visit day, enumerators recorded real-time attendance of both students and teachers and compared it with administrative attendance records to assess consistency. They also observed and logged teacher punctuality and classroom presence to evaluate alignment with official reporting. These surprise checks helped ensure data integrity and provided insights into the day-to-day functioning of the classrooms beyond reported figures.

Announced visits & midline (TBD)

During the announced visits and midline assessment, conducted between September and October 2022, researchers observed a 60-minute classroom session with the sample math teacher in all 468 schools to evaluate instructional practices. Following the observation, teachers completed a 40-minute survey to share their experiences and perceptions regarding the intervention and teaching conditions.

The team then revisited the same 20 baseline students in each school for follow-up assessments. Each student completed a 15-minute survey (in all 468 schools) aimed at tracking changes in their learning environment. In the 312 treatment schools, these students also participated in a 30-minute group-based written assessment to measure the learning impact of the interventions.

All survey and assessment responses were recorded on paper and later digitized. A dedicated in-house data entry team handled this task, ensuring the timely and accurate digitization of the midline data across all school visits.

Endline Survey

In October and November of 2022, the study entered its endline phase. The 20 Grade 6 students tracked from the baseline were now assessed through multiple surveys and evaluations across all 468 participating schools.

First, a 15-minute digital student survey was conducted individually with each student. Following this, a 180-minute group-based written assessment covering both math and reading was administered on paper. This was facilitated by the school's math teacher, with data entry to be completed afterward. Finally, students completed a 45-minute paper-based survey collecting further information. These three components—survey, assessment, and follow-up survey—were implemented consistently across all 468 schools.

In addition to these core activities, a range of field data collection efforts were implemented during the endline phase:

  1. Teacher and Student Attendance Data (August to October) were collected in all schools.
  2. Classroom Observations, Interaction Monitoring, and Teacher Records were maintained across control, Treatment 1, and Treatment 2 groups.
  3. A Spot Checker team (1 per division) was deployed due to time constraints, replacing backchecks. These spot checkers collected:
               -Teacher and Student Attendance
               -Seating Data  
               -Interaction Records
               -Observed Teacher Surveys, when feasible

To address prior data quality issues from the baseline:

  1. A Back Check was introduced specifically for the Impact Test data.
  2. As a result, two data entry operators were suspended and their entries redone.
  3. Double data entry was not possible due to budget limitations.
Further data integrity checks included:

  1. A mismatch analysis of student Interaction Records and Seating Data, showing a very low error rate (e.g., only 12 of 366 repeat-recorded students had internal mismatches; 6 out of 840 matched students were misrecorded as absent)
  2. Several data issues were noted and labeled in the datasets, including:
                 -Students returning after long absences
                 -Lost teacher and student attendance records in a few schools
                 -Classroom sections that were split mid-year

Training for field staff involved two days of Zoom-based instruction supported by user manuals, followed by one day of in-person training near Mirpur DOHS to practice interaction and snapshot protocols. These comprehensive activities marked the conclusion of the

Engagement:

Sharnic Djaker, Alejandro J. Ganimian, and Shwetlena Sabarwal published the study Out of Sight, Out of Mind? The Gap Between Students’ Test Performance and Teachers’ Estimations in India and Bangladesh: examining the mismatch between students' actual test scores and teachers' estimations in India and Bangladesh.






image not available

OVERVIEW

TOPICS

 Teacher Training, Low-Achieving Students, Formative Assessment, Educational Intervention, Bangladesh

LOCATION

12 Districts in Bangladesh Rajshahi Division: Nawabganj, Joypurhat, Sirajganj, Naogaon Mymensingh Division: Mymensingh, Netrakona, Jamalpur, Sherpur Chattogram Division: Comilla, Brahmanbaria, Lakshmipur, Rangamati

TIMELINE
 February 2022 –  March 2023

SAMPLE SIZE

Total Schools: 468 (156 control, 156 Treatment 1, 156 Treatment 2)  
Students: Approximately 21,757 Grade-6 students across all schools.  
Teachers: 463 Grade-6 math teachers surveyed


PARTNER

 ARCED Foundation, Secondary Education Development Program (SEDP), Teach for Bangladesh

PERSONNEL
 Noam Angrist, TM Asaduzzaman, Sharnic Djaker, Alejandro Ganimian, Shwetlena Sabarwal,  ARM Mehrab Ali, Sadia Sumaia Chowdhury, Zahirul Islam, Tasnim Binte Mamun, Fahmidur Rahman, George Mithun Roy, Hasibur Rahman Sohan

PUBLICATIONS

Out of Sight, Out of Mind? The Gap Between Students’ Test Performance and Teachers’ Estimations in India and Bangladesh

NEWS


Team

Projects

About Us

Career

Twitter

Facebook

LinkedIn

GitHub

ARCED Foundation

Aureolin Research, Consultancy & Expertise Development Foundation
House 463, Flat 5A & 6A, Road 7, Avenue 5
Mirpur DOHS, Dhaka 1216, Bangladesh
info@arced.foundation
+880 251 051 088

© Copyright 2022. All Rights Reserved