An independent project about PISA in Kazakhstan

PISA is not a ranking. It is a mirror

Every three years, 15-year-olds in 91 countries show whether they can use what they know in unfamiliar situations. Here we take an honest look at Kazakhstan’s results and ask what should change in the classroom.

PISA 2025 results For teachers For parents

Where Kazakhstan sits on the reading literacy scale

PISA 2025 · mean score

Level 2

Kazakhstan
385

461
OECD average

Level 2 is the baseline: students understand a text and can identify its main idea. 36% of students in Kazakhstan reach it in reading.

PISA 2025 in numbers

Science holds steady, mathematics declines, reading has not moved in 16 years

The OECD released the first results of the ninth cycle on 8 September 2026. The full analytical package is expected in May 2027.

22,680

students from 643 schools in every region took part

420 / 482

points in science, the main domain of this cycle; after the slash, the OECD average

414 / 463

points in mathematics, down from 425 in 2022

449 / 500

points in the new domain, Learning in the Digital World

Reading: 385 points in 2025, almost the same as in 2009

* Kazakhstan’s 2015 results were judged not comparable. Source: OECD, PISA 2009–2025.

Reading is the main gap

36% of students reach the baseline. Reading underpins every subject: a student who misreads a problem cannot solve it in physics either.

Very few top performers

About 1% reach the highest levels (5–6) in mathematics. The system needs to lift weaker students and grow strong ones at the same time.

A strength: equity

The science gap between students from the most and least advantaged families is 41 points, against 85 on average across the OECD.

About PISA

What PISA measures and how it differs from the UNT

PISA does not test the curriculum. It tests whether students can apply knowledge: understand a text, work with data, reason and draw conclusions.

The Programme for International Student Assessment is run by the Organisation for Economic Co-operation and Development (OECD). Since 2000 it has assessed 15-year-olds every three years, as they approach the end of compulsory schooling.

Not every school takes part. PISA uses a scientifically drawn sample that describes the education system as a whole, not individual students or teachers. Each student answers only some of the items, so there are no individual PISA scores.

Kazakhstan has taken part since 2009 and has completed six cycles: 2009, 2012, 2015, 2018, 2022 and 2025.

Reading

Locate information, understand meaning, judge how reliable a source or an argument is.

Mathematics

Turn a real-life situation into a mathematical problem, solve it and check the answer.

Science

Explain phenomena, design a way to test a hypothesis, interpret data. The main domain in 2025.

Innovative domain

Changes every cycle. In 2025: Learning in the Digital World.

Questionnaires

Students and principals describe their school, lessons, families and use of digital devices.

	PISA	UNT (national university entrance test)
Purpose	Monitoring the education system	An admission decision for each graduate
Who takes it	A sample of 15-year-olds	Everyone applying to university
What it measures	Applying knowledge in unfamiliar situations	Mastery of school subjects
Stakes for the student	None; results are not reported	A scholarship and a university place
How often	Every three years	Every year

Can students prepare for PISA?

Drilling for PISA does not work: the items are new each cycle and the sample is random. What works is teaching students to understand texts, reason and work with data in every lesson. That is what the results reflect.

Why doesn’t every school take part?

A representative sample is enough to describe the system. In 2025 in Kazakhstan it covered 643 schools across all regions.

Who will see my child’s results?

No one. PISA does not report individual scores: the data are anonymous and analysed only at the level of the country and groups of students.

Why are Nazarbayev Intellectual Schools not in the national result?

NIS take part as a separate sample. The OECD reports their results separately and does not mix them with the national figure.

Where can I read the official reports?

On the OECD website: oecd.org/pisa. It also has an interactive dashboard with data for every country.

For teachers

Task of the week

Each week: one task in the spirit of PISA, with notes on what it assesses and how to use the same approach in your own lesson.

Reading literacy · multiple sources Original teaching task, not from the OECD item bank

Source A · school newspaper

Time to take phones out of lessons

The principal of School No. 12 proposes collecting phones before the first lesson from January. “Students are constantly distracted,” he says. “At break nobody talks to each other.” The parents’ committee supports the idea.

Source B · survey of 240 students in grades 8–10

What do you use your phone for in class?

Looking up information for a task58%

Messaging, social media41%

Using an AI chatbot33%

I don’t use it12%

Students could choose more than one answer. Illustrative data.

Question 1

Which source gives an opinion and which gives data? Give one example from each.

Question 2

Aigerim writes: “The survey proves that phones get in the way of learning.” Do you agree with her? Support your answer with evidence from both sources.

Notes for teachers

What it assesses: whether students can judge if the data support a conclusion. In PISA this is the “evaluate and reflect” process; tasks like this usually sit at Levels 4–5.

Full-credit answer: disagree, or agree only in part. The survey shows that 41% get distracted, but 58% use their phones for learning. The data support the claim only in part.

Common mistake: the student agrees by repeating the principal’s words from Source A and ignores the numbers.

In your lesson: give students any claim from a textbook or the news and ask what data they would need to check it.

Proficiency levels in plain language

Six levels for each domain, with examples of what a student can do at Level 2 and at Level 5.

Reading in every subject

Short techniques for mathematics, physics and history teachers: working with a problem statement, a graph, a source.

AI in the classroom

80% of students in Kazakhstan say they evaluate AI answers in class. How to turn that into a critical reading skill.

Five-minute videos

Walkthroughs of released OECD items in Kazakh and Russian.

For parents

What families can do

PISA results do not depend on school alone. Three things that work at home and need no tutor.

54%

Talk about school

The share of students whose parents discuss school with them every week fell from 68% to 54% in three years. Ask “what was hard today?” instead of “what grade did you get?”

36%

Read together

Only a third of teenagers read at the baseline level. Fifteen minutes a day discussing an article or a book (what did the author mean, do you agree?) trains exactly what PISA measures.

39%

Agree on AI

That is the share of students who use chatbots when writing assignments. AI helps when it checks a child’s thinking and harms when it replaces it. Ask them to write their own version first.

Analysis

Short briefs for decision-makers

Four pages, one question, OECD data and conclusions for Kazakhstan.

January 2027

Why reading is not improving: 16 years of data

What has changed in schools since 2009, and what has not.

In preparation

March 2027

AI, chatbots and homework

How students in Kazakhstan use AI and what it means for reading.

In preparation

April 2027

Where are our top performers?

About 1% at the highest levels in mathematics: causes and international experience.

In preparation

May 2027

The OECD’s full PISA 2025 report

A review in three languages on release day.

Tracking

Author’s column

Five points down in sixteen years

In 2009, 15-year-olds in Kazakhstan scored 390 in reading. In 2025, they scored 385. In between, curricula and textbooks changed, and digital platforms and an updated curriculum arrived. The reading score barely moved.

The pandemic cannot explain this: the plateau began long before 2020. Nor can a lack of resources: according to PISA 2025, schools are better equipped and students rate teacher support highly.

That leaves the key question: what happens in the lesson? Do students read complex texts, argue with the author, compare sources? Or is reading still seen as the literature teacher’s job? The first brief, in January, will look at this.

About the author

Miras Baimyrza

Head of the AIT and Artificial Intelligence Department at the National Testing Centre of Kazakhstan. Leads the development of the Admissions Insight Test (AIT), Kazakhstan’s next-generation university admissions test, built in partnership with ETS.

More than ten years across the full assessment cycle: national monitoring (MODO), international studies (PISA, TIMSS, PIRLS, TALIS, ICILS, PIAAC), and test development and validation. Represented Kazakhstan on the PISA and TALIS Governing Boards and led research at Nazarbayev Intellectual Schools, including the annual NIS Conference.

Graduate of the Nazarbayev University Graduate School of Education. On this site he writes about what international assessment data show and how to turn them into decisions for schools and classrooms.

PsychometricsLarge-scale assessmentAI in assessmentEvidence-informed policyResearch capacity building

LinkedIn profile