Learning & Skills Brief — November 2026

This month the results depended on what actually reached learners, more than on the type of programme. That meant the information teachers had about their pupils, how often coaches observed teachers, and how a tool was used in the course. In Bangladesh, cheap test reports changed what teachers did in class and raised pupils' scores. In the United States, a large coaching programme and a university AI tutor produced no gain, and in one case a loss, once that contact thinned out. In East Africa, vocational training systems are moving training and assessment into the workplace.

In Bangladesh, giving teachers two short test reports raised maths scores for about $5 per pupil

• Source: What Works Hub for Global Education, working paper 2026/003 (Djaker, Ganimian, Angrist & Sabarwal) · 21 October 2026 · wwhge.org
• The intervention: grade-6 maths teachers in 312 schools in Bangladesh received the results of two short tests, reports flagging the topics and pupils that needed support, and an annotated answer key.
• Baseline: teachers could not estimate their pupils' scores, rank or spread, and were twice as likely to interact with high-achieving as with low-achieving pupils.
• Change in teaching: teachers lowered their estimates of pupils' performance, spread their attention more evenly and spent more time on foundational content. The flagged topics and pupils improved most.
• Results: on independent end-of-study tests, maths rose by 0.23 standard deviations, reading by 0.17 and three social-emotional skills by 0.10 to 0.14. Gains were similar across initial levels. The cost was $5.06 per pupil.
• Why it matters: When teachers know little about where each learner stands, giving them that information in a usable form changes whom they teach and what they teach. It is cheaper than adding training hours. Caveat: this is a working paper, and the study's duration and the details of its comparison groups are not in the summary.

A large teacher-coaching programme lost its effect as it grew

• Source: Education Week, reporting a study in Educational Evaluation and Policy Analysis led by Michelle Doughty (Tennessee Education Research Alliance) · 27 October 2026 · edweek.org · study: journals.sagepub.com
• The programme: an instructional coaching programme in a large US urban district grew between 2014 and 2022 to nearly 1,000 mentor teachers and more than 6,000 teachers supported, rolled out to schools in five waves. Education Week calls it the largest evaluation of a coaching programme to date. Researchers compared participating schools before and after they joined with similar schools that did not join; the design was not randomised.
• Results: no effect on teacher retention or on pupils' state test scores in maths and reading, although the first schools to join showed early promise.
• How the dose fell: as the programme expanded, the number of teachers to coach outgrew the supply of mentors. Observations fell from five or six per teacher per year to one or two. Mentors also taught half-time.
• Why it matters: In coaching, the number of observations and feedback sessions each teacher gets is what drives results, and that number falls first when a programme scales. Anyone scaling a coaching or mentoring scheme should set a minimum coach-to-teacher ratio and number of observations, and track them as the programme grows. Caveat: the full paper was not available; this summary relies on press coverage, which reports no effect sizes.

Adding an AI tutor to university courses lowered grades in a randomised trial

• Source: Liu et al., EdWorkingPaper 26-1598, Annenberg Institute (Brown University) · October 2026 · edworkingpapers.com · context: Inside Higher Ed · 7 October 2026 · insidehighered.com
• The trial: at the University of Maryland (United States), 2,379 undergraduates and 30 instructors in several disciplines took part in the autumn 2025 semester. Course sections were randomly given access to a generative AI tutor built into the learning platform.
• Results: among sections of the same course, access reduced final grades by 0.37 standard deviations and participation on the learning platform by 0.90 standard deviations. Losses were larger for first-generation students.
• Use of the tool: about 15% of students used it. 73.8% of requests asked for information, explanations or solutions, and 0.7% asked for feedback on the student's own work. By default the tool gave answers directly; few instructors switched it to the mode that guides students toward the answer, and instructors received light training.
• Why it matters: Adding an AI tool to a course without redesigning the activities can draw learners away from the practice that produced learning. Before deploying one, decide whether it gives answers or guides, train the facilitators, and measure participation before and after. Caveat: one university and a working paper; the university itself says the findings are not clear-cut.

Rwanda brings trade associations in to host vocational trainees in the workplace

• Source: Rwanda TVET Board (RTB) · 15 October 2026 · rtb.gov.rw
• The agreements: RTB signed memoranda of understanding with four associations: the Rwanda Garage Association, Rwanda Wood Value Chain Association, Rwanda Tailors Association and Rwanda Hospitality Association.
• What they cover: dual training, workplace placements, internships, apprenticeships and industrial attachments. They also cover training for workplace mentors and in-company instructors, and industry input into occupational standards, curricula, assessment and quality assurance.
• Scale: the associations give access to more than 1,680 businesses and about 1,500 individual tailors. No number of trainees, places or budget is given.
• Why it matters: Working through associations rather than firm by firm makes it cheaper to find placements. The constraint then becomes the number of trained workplace mentors, and their training needs its own budget line. The number of placements actually opened is the figure to watch.

Uganda moves vocational diploma assessment to 70% practical work

• Source: Nile Post (Kampala), report of a stakeholder workshop at the Uganda Vocational and Technical Assessment Board (UVTAB) · 8 October 2026 · nilepost.co.ug
• Assessment: at diploma level, practical work will count for 70% of the assessment and theory for 30%. UVTAB says training and assessment will be based on actual workplace requirements, with employers more involved in curricula, training and assessment.
• Trainers and institutions: trainers in public and private institutions are to be reskilled or upskilled. Registering an institution will require suitable infrastructure, equipment and qualified trainers, followed by compliance reviews.
• Sector committees and tracking: the TVET Council has set up 15 sector skills expert committees, covering agriculture, manufacturing, tourism, business, sport, oil and gas, and minerals, among others. The council will also track graduates' employability. No start date is given.
• Why it matters: Giving practical work more weight raises the cost per candidate (equipment, workshops, assessor time) and changes the profile of trainers needed. Programmes that prepare candidates for Ugandan diplomas will need to plan for practical assessment conditions, not only content. Caveat: a single press source, with no date of entry into force.

Rwanda plans a national education data platform, with teacher feedback on textbooks before printing

• Source: Top Africa News, reporting Education Minister Joseph Nsengimana's presentation to Parliament on 20 October · 26 October 2026 · topafricanews.com
• What it is: the Ministry of Education is preparing an "Education Intelligence Center", a platform for data on pupils, teachers, schools, classrooms, learning materials and performance indicators.
• Intended uses: flag pupils at risk of dropping out (including frequent absences), detect teacher shortages, overcrowded classrooms and poor results, guide resource allocation and assess literacy and numeracy policies.
• Teachers and textbooks: according to the Director General of the Rwanda Basic Education Board, school leaders will report which textbooks they hold and how old they are. Teachers will be able to comment on textbooks before they are printed and distributed. Head teachers will have to submit data on their schools regularly.
• Timeline: the project is being prepared with the Ministry of ICT and the World Bank. A developer is due to be selected by March 2027, followed by six to seven months of development.
• Why it matters: Asking teachers to review materials before printing is a cheap step that any programme producing learning materials can copy. For organisations working with Rwandan schools, national indicators should become available from late 2027, and aligning their own monitoring indicators with them now would avoid double reporting. Caveat: a single press source and no cost figure.