In this article
The OECD published its PISA results on 8 September 2026, and the headline was that reading among 15-year-olds has reached its lowest point since the study began. Across OECD countries, reading scores dropped 28 points over the past decade, which the OECD puts at about a year and a half of learning.
UK students did better than that suggests, averaging 494 in reading and 488 in maths against OECD averages of 461 and 463, with England’s scores holding steady since 2022. The interesting number for a UK parent is a different one.
PISA also measured what it calls hasty reading, where a student moves through a text at speed and gets it wrong. That behaviour has nearly doubled since 2018, and it is the finding that reaches into maths.
1. Most of a Maths Question Is Reading
Take what a iGCSE question actually requires. Read a scenario. Work out which numbers matter. Notice which ones are there to mislead. Decide what is being asked for. Choose a method. Carry it out. Present the answer in the form requested.
One of those seven is calculation.
Most secondary students can read the words on the page perfectly well. What has changed is the default speed. Years of scanning trains a reader to take the gist and move, which serves a group chat and ruins a multi-step problem, where every clause is load-bearing. Reduced by 20% and reduced to 20% differ by one word and about twenty pounds.
2. Five Ways to Lose a Mark You Had Earned
| The question | What the student wrote | What it wanted |
|---|---|---|
| A jumper costs £40 and is reduced by 20% in a sale. Find the sale price. | £8 | £32. The percentage was calculated correctly, then handed in as the answer |
| Anna buys three packs of twelve pens and gives eight away. How many are left? | 36 | 28. Stopped one step early |
| A rectangle is 8cm by 6cm with a diagonal of 10cm. Find the perimeter. | Pythagoras working | 28cm. The diagonal was there to be ignored |
| Which of these is not a multiple of 6? | 18 | The one that isn’t. Four letters, whole mark |
| Give your answer in centimetres. | 1.4 | 140 |
None of these is a maths error. Every one of them costs the same as a maths error, which is what makes them so annoying to a child who knew how to do it.
3. It Gets Worse as the Maths Gets Harder
Plenty of children sail through primary arithmetic and hit trouble at Key Stage 3, without the arithmetic having got harder. The questions got wordier.
Three things arrive at once in secondary maths. Problems go multi-step, so an intermediate result has to be held while working towards something else. Questions start including numbers that are not needed. And the form of the answer gets specified, in units, in decimal places, or in a demand to show that something is true.
Each of those is a reading task before it is a mathematical one. A student fluent in every technique on the syllabus can still bleed marks across a paper.
4. Telling a Gap from a Rush
This distinction changes what you do next, and it takes ten minutes at the kitchen table.
Take a question they got wrong and give them the bare calculation with the words stripped out. If they can do it, the maths is intact and the mark went to reading or attention. If they cannot, there is a genuine gap and more worded practice will not touch it.
Then look at the shape of the errors across a few pieces of work. A knowledge gap is topic-shaped and consistent. The same kind of question fails every time, calm or rushed. Rushed processing scatters across unrelated topics, the work is usually nearly right, and it improves the moment the same question is asked again slowly.
One more signal. Frustration when they see the mark points to attention, because they know how to do it. Confusion points to a gap.
5. What Helps at Home
Before any calculating, three questions out loud. What has the question given me. What is it asking for. Is there anything here I do not need. Then one at the end: is that answer sensible. A speed of 400mph or a discount bigger than the original price gets caught here.
Underlining the quantity being asked for works well. So does crossing out the numbers that turn out to be decoration.
Attention responds to a protected window more than to a long session. Thirty minutes with the phone in another room beats three interrupted hours. Ask what they learned rather than whether the homework is done, since explaining is a retrieval task and being finished is not.
Reading for enjoyment has been falling across the OECD and it is the habit most closely tied to the skills that dropped. The format matters less than the length. Sustained attention on one text is the thing.
6. Screen Time and AI, Honestly
PISA found digital distraction associated with lower science scores in 59 of the 82 systems studied. It also found that students using AI chatbots for drafting, summarising and researching scored around 20 points lower in science than students who did not.
Both are associations, from a snapshot rather than an experiment, and the AI figure is in science, not maths. Students already struggling are more likely to reach for something that removes the difficulty in the first place. Estonia is the useful check on any tidy story here: heavy device use, no significant link to lower performance, and one of the strongest systems in the study.
The OECD did not recommend phone bans, and the evidence does not carry a single household policy. What it does support is protecting the stretch of time where the work happens.
7. Finding Out Which One It Is
If the kitchen-table test left you unsure, a Full Curriculum Assessment on Ivy Maths goes through the year group topic by topic and marks off what is secure, naming what is not. It answers the question a school report cannot, which is whether the marks are going to maths or to reading.
For the day-to-day version, Check My Homework shows where the working went off, which is a different thing from handing over the answer. A student who finds their own misread question learns something. A student handed £32 learns nothing.