Back to History Current Version

Individualized Self-learning Program to Improve Primary Education: Evidence from a Randomized Field Experiment in Bangladesh

Last registered on April 23, 2018


Trial Information

General Information

Individualized Self-learning Program to Improve Primary Education: Evidence from a Randomized Field Experiment in Bangladesh
Initial registration date
April 21, 2018

Initial registration date is when the trial was registered.

It corresponds to when the registration was submitted to the Registry to be reviewed for publication.

First published
April 23, 2018, 7:23 PM EDT

First published corresponds to when the trial was first made public on the Registry after being reviewed.



Primary Investigator

Faculty of Economics, University of Tokyo

Other Primary Investigator(s)

PI Affiliation
Bangladesh Institute of Development Studies
PI Affiliation
JICA Research Institute
PI Affiliation
University of Tokyo
PI Affiliation
University of Tokyo

Additional Trial Information

Start date
End date
Secondary IDs
This paper reports on the results from a field experiment that tests the effectiveness of the globally popular Kumon learning method in improving the cognitive and non-cognitive abilities of disadvantaged pupils in Bangladesh. Using a randomized control trial design, we study the impact of this individualized self-learning approach among third and fourth graders studying at BRAC non-formal primary schools. The results show that students of both grades in the treatment schools record substantial and significant improvement in their cognitive abilities as measured by two different mathematics tests (Kumon diagnostic test score per minute and proficiency test score) after a period of 8 months, compared to students in the control schools. In terms of non-cognitive abilities, the results give some evidence of positive and significant impacts, particularly on the self-confidence of the pupils. Interestingly, this intervention also had a positive and significant impact on the ability of teachers’ to assess their students’ performance. Overall our results suggest the wider applicability of a properly designed non-formal education program in solving the learning crisis in developing countries.
External Link(s)

Registration Citation

Kawarazaki, Hikaru et al. 2018. "Individualized Self-learning Program to Improve Primary Education: Evidence from a Randomized Field Experiment in Bangladesh." AEA RCT Registry. April 23.
Former Citation
Kawarazaki, Hikaru et al. 2018. "Individualized Self-learning Program to Improve Primary Education: Evidence from a Randomized Field Experiment in Bangladesh." AEA RCT Registry. April 23.
Sponsors & Partners

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information
Experimental Details


In Bangladesh, Kumon Method of learning has been introduced in selected BPS among the third and fourth grade of students as a supplementary module in mathematics. Kumon aims to enable students to develop advanced academic and self-learning ability by ensuring that children are always studying at a level that is just right for them. In particular, students are assigned to an initial level based on individual performance in a diagnostic test (DT) provided by Kumon Institute, not on the basis of the school grade or age. Kumon Method is uniquely designed to set the initial level slightly lower than the concurrent maximum capacity of the student in order to: i) ensure full understanding of the basic concepts as a firm building block of cognitive ability development; and ii) stimulate students’ motivation to continue to study, which also works for the development of their non-cognitive ability such as self-confidence and self-esteem. Kumon’s mathematics program is divided into 20 levels (from Level 6A to Level O), and five elective levels, comprising a total of 4,420 double-sided worksheets. All of these worksheets are carefully designed, starting from simple counting to advanced mathematics, with the level of difficulty increasing in small steps.8 Worksheets contain example questions with hints, which help students to acquire step by step problem solving skills by themselves. As a result, students can absorb material beyond their school grade level through self-learning, and advance to studying high school level material at an early age. Importantly, slower learners can spend more time on the basics without being rushed to move on to advanced level of materials beyond their level of understanding.

Another feature of Kumon Method of Learning is a tracking system of each student’s progress and achievements using personalized record books. Kumon instructors do not teach in the class and hence, do not need extensive prior experience to conduct daily quizzes to monitor the understanding and progress of each student. This is because Kumon worksheets are laid out in small steps to enable students to self-learn and there is a determined standard time per worksheet to judge whether students can advance to the next level or should repeat a level. Having detailed progress reports on the worksheets allows instructors to obtain more objective information about their students’ abilities, and their understanding of the mathematics involved.
Intervention Start Date
Intervention End Date

Primary Outcomes

Primary Outcomes (end points)
(1) Cognitive outcomes such as achievement test scores from the Kumon method of learning and nation-wide public examinations; (2) non-cognitive outcomes such as self-esteem, grit, and other psycho-social measures; and (3) teachers' assessment skill of students.
Primary Outcomes (explanation)

Secondary Outcomes

Secondary Outcomes (end points)
Secondary Outcomes (explanation)

Experimental Design

Experimental Design
The intervention is a randomized controlled trial with one treatment and one control arm.
Experimental Design Details
We randomly selected 34 BPSs having third and fourth graders from the 179 BRAC nonformal primary schools in Dhaka and surrounding areas, whereby 17 schools received Kumon materials and 17 schools did not receive these materials so that they could serve as treatment and control schools, respectively.10 The resulting sample breakdown by grades is as follows: 19 (out of 48 schools) for the third grade and 15 (out of 131 schools) for the fourth grade are tracked in our study.11 In total, our study started tracking roughly 1,000 students in these 34 schools. In these schools, we choose only one of the two class shifts (either morning or afternoon), with an average class size of 30 students. The intervention for the treated school students consists of a 30 minute session on Kumon study prior to the beginning of their regular lesson. Thus, students in the treatment schools come to school earlier than usual during the experiment periods.
Randomization Method
Randomization has been done in office by a computer using a random number generator.
Randomization Unit
School level randomization
Was the treatment clustered?

Experiment Characteristics

Sample size: planned number of clusters
34 schools
Sample size: planned number of observations
1,000 pupils
Sample size (or number of clusters) by treatment arms
17 schools treatment, and 17 schools control
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
0.4 standard deviation

Institutional Review Boards (IRBs)

IRB Name
The research ethics committee at the University of Tokyo
IRB Approval Date
IRB Approval Number


Post Trial Information

Study Withdrawal

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information


Is the intervention completed?
Intervention Completion Date
April 13, 2016, 12:00 +00:00
Data Collection Complete
Data Collection Completion Date
January 31, 2018, 12:00 +00:00
Final Sample Size: Number of Clusters (Unit of Randomization)
34 Schools
Was attrition correlated with treatment status?
Final Sample Size: Total Number of Observations
Final Sample Size (or Number of Clusters) by Treatment Arms
17 schools control, and 17 schools treatment
Reports, Papers & Other Materials

Relevant Paper(s)

This study investigates the effectiveness of a globally popular method of self-learning at
the right level in improving learning outcomes—the cognitive and noncognitive abilities
of disadvantaged students—in a developing country, Bangladesh. Using a randomized
controlled trial design, we find substantial improvements in cognitive abilities measured by
math test scores and catch-up effects in terms of noncognitive abilities or personality traits
measured through a self-esteem scale. Moreover, our study is the first to use alternative
cognitive ability measures, that is, time reduction as well as time-adjusted test score,
which are critical dimensions of cognitive development. Subsequently, we investigate
the long-term effects using students’ math results of the national-level exam. We find a
reasonable longer-term impact on cognitive abilities 20 months after the intervention for
younger students. Our estimates indicate that the program’s benefits exceed its costs.
Yasuyuki Sawada, Minhaj Mahmud, Mai Seki, and Hikaru Kawarazaki (2023), "Fighting the Learning Crisis in Developing Countries: A Randomized Experiment of Self-Learning at the Right Level," accepted at Economic Development and Cultural Change.
This study investigates the effects of speed competition in classrooms on young pupils' learning outcomes. To examine how faster peers' speed affects slower pupils' speed and learning, we employ students' daily progress data in a self-learning programme at BRAC primary schools in Bangladesh. The programme's unique setting allows us to address the reflection problem reasonably well. While speed competition could generate negative consequences, our results show overall positive peer effects on problem-solving time and scores. The effects are stronger among peers with similar abilities, without negatively affecting others. Our results show efficiency gains from non-market competition in education and learning.
Kawarazaki, H., Mahmud, M., Sawada, Y. and Seki, M. (2023), Haste Makes No Waste: Positive Peer Effects of Classroom Speed Competition on Learning. Oxford Bulletin of Economics and Statistics, 85: 755-772.

Reports & Other Materials