Phonics in primary schools: an unfinished revolution

Since 2011 schools have been testing children in year one (age 5/6) on their phonics ability. The test involves reading about 40 words. There are some real words and some made-up ones, because the goal is to test pupils’ ability to decode the words and put sounds together.

The test is to identify which children need more help in learning to read, and was introduced as part of the Gove / Gibb drive to get schools to adopt phonics - the most effective way to help children to read.

As well as the test, the new government provided funding to 14,000 schools for phonics training and resources.

Phonics is now well established in England, though amazingly in Labour-run Wales the government continues to push a discredited method called “cueing”. This means that rather than learning letter sounds and how to blend them together, children in Wales are shown whole words along with pictures (the ‘cue’) in the hope they will memorise the whole word. The result has been a complete disaster.

In contrast, in England, the shift to phonics has seen a much larger number of pupils mastering the basics of reading in year one, which is one reason England’s schools have risen up the international league tables, particularly in comparison to Scotland and Wales. The NEU continues to object to the test on strange ideological grounds, and indeed campaign for “the removal of all statutory testing in primary schools”. Though a bit over a third of teachers would like the check scrapped, they are in the minority.

Here is the overall share of pupils who have hit the Phonics standard over time. There were big improvements for everyone - both those on Free School Meals (FSM) and those who were not.

Encouragingly, this improvement in reading in year one was followed by improvements in reading ability at the end of primary school (Year 6). New National curriculum standards were introduced in 2015/16, so the series breaks there, but we can see that:

• The percentage of pupils meeting the expected standard in reading at the end of Key Stage 2 (end of primary school) rose from 66% in 2015/16 to 74% in 2023/24

• The percentage of pupils meeting the higher standard in reading at the end of Key Stage 2 (end of primary school) rose from 19% in 2015/16 to 29% in 2023/24

So rising standards at the start of primary school led to rising standards in the exams at the end.

However, while there has been real progress since 2010:

- Progress was battered by the pandemic. Though still much higher than 2010, as of last year the proportion of pupils meeting the expected standard was still just returning to its pre-pandemic peak.

- There is a big gap between richer and poorer pupils. Although this was closing somewhat from 2010 to the pandemic, it remains large.

- Among some groups, and in some places, fewer pupils are meeting the standard.

This post looks at those variations and how things have changed over time.

Learning to read - progress so far

Why does this matter? Gaps in capability at the start of school tend to snowball. Children who feel left behind become demoralised and struggle to keep up. Knowledge is cumulative: the more you know, the easier it is to learn more. The educationalist E.D. Hirsch likened knowledge to strips of Velcro – one piece of knowledge snags and helps pull in another.

How powerful is this effect? Hugely. Children born in August are 20% less likely to go to a Russell Group University than children born in September. These two groups of children are alike in every way, except that the August babies were the smallest in their year at the start of school, and so started from behind.

For this reason the world’s best performing educational systems - mainly in Asia – have an obsession with making sure everyone gets the basics: the result is high performance and few left behind. Asian systems are demanding and aspirational, but actually quite egalitarian in their effects.

In England we still see too many groups and places where the phonics standard is not being met.

The table below looks at how the proportion who meet the standard varies by FSM, sex and ethnicity.

As the table below shows, pupils from a Chinese background outperform everyone, whether on FSM or not. But they are a small group. Looking at bigger groups, pupils from a Indian background who are on FSM outperform pupils from a black background who are not.

But as the table below shows, ethnicity and income interact in different ways. While white British pupils who are not on FSM are in the middle of the pack when it comes to pupils who are not on FSM, white British pupils who are on FSM are less likely than any other group to meet the standard, which is pretty astonishing.

The worst result of all is found among white boys on FSM, with only six in ten meeting the standard¹.

The FSM ‘gap’ is much bigger for White British kids than any other group. The gap is 2% for pupils from an African or Bangladeshi background, 5% for Pakistani, 6% for Caribbean or Indian background kids, but 20% for White British.

The male-female gap is much bigger for black children and white British kids on FSM than other groups. For Black Caribbean kids on FSM the difference is 80% for girls vs 66% for boys. For White British pupils on FSM it is 71% vs 60% - an incredible demonstration of the power of culture.² Perhaps we need to rewrite Harry Potter with Ron as the swot.

As the twig is bent

How powerful a predictor of future attainment is the phonics test? The chart below shows how each of the groups above did in year one and in GCSEs in year 11. Even just taking a one year snapshot and looking at whole groups the link is strong - I suspect if you used the National Pupil Database you would find an even stronger link between the individual’s performance at year 1 and year 11. That is pretty striking, given that the year 1 test lasts about ten minutes and involves reading 40 words.

That said, there are variations - among those not on FSM, white British and black Caribbean kids do worse at GCSE than you might have expected, and Indian and Bangladeshi kids do better.

Where are pupils doing better?

Looking around the country we see big variations. The interactive map below makes the size of each local authority proportional to the size of its population.

Overall, we see the largest proportions meeting the standard in North and West London, and affluent suburbs like Trafford, Solihull, and Cheshire East. The home counties do fairly well. Lots of other cities do badly: Leicester, Nottingham, Manchester, Liverpool. And yet there are other urban areas like Barnsley and Wakefield that are doing well.

It is not too surprising that affluent areas should do better.

But if we look at the results for those on free school meals the story is different – London still does well. But East London also does well for FSM kids (showing that their total above is quite low because so many are on FSM).

Wakefield and Barnsley do better, but so does Birmingham. In contrast in a lot of lower-funded shire areas (particularly around London) the Phonics scores for children on FSM are low. The area around Bristol is also bad.

Is this a tribute to schools and early years settings in London? An underlining of the vast funding gap between the capital and the shires? Perhaps – though it is also surely entangled with the demographics. Many things about London schools can be explained by its demographics.

As noted above, among children on FSM, the white British are far and away the lowest scoring, and white British children are a very low share of pupils in Greater London - just over one in five (22.7%) compared to just over 60% nationally.

But then that still doesn’t mean they aren’t doing something right in the capital. It’s not inevitable that ethnic minorities should do better - in fact the reverse is true across most of the OECD³. Demographics also don’t explain the better results in places Barnsley and Wakefield, or Redcar or Middlesborough or Warrington.

To properly untangle this we would need school data, or access to the National Pupil Database.

For good reasons school level data is not published. In fact the Phonics check is a good example of something we need more of: “low stakes testing” - it’s there to see which children need help, rather than as a school accountability tool.

That said, DFE should make it much easier to access the National Pupil Database, which has been made so difficult to access that many high-quality researchers can’t get hold of it. (One centrist think tank I know is currently building a “special room” so they can get access to it - don’t ask.)

Conclusions

The year one Phonics check is the first “proper” test of how well pupils are doing, and quite a powerful predictor of later attainment.

Yet strangely the only target the Labour government have set themselves in education is a target for a much softer measure, the Early Years Foundation Stage (EYFS) school readiness measure.

This is a much softer, earlier measure, self-assessed by nursery teachers. The things assessed are things like “building relationships” and “being imaginative”.

Labour’s target is to have 75% of five-year-olds in England reach a "good level of development" on this measure by 2028. Interestingly, girls are already reaching this standard (exactly 75% in 2023/24), so the change that is needed is among boys (of whom just 60.7% are assessed as hitting it).

Interestingly, the female-male gap is twice as big in EYFS than in the year 1 phonics check (75% - 61% vs 84% - 77%) which may reflect the nature of what is being assessed.

There is nothing wrong with the early years framework (it’s age-appropriate) but it is inevitably more subjective. And when you combine a government target with a fairly subjective measure there is a high probability that Goodhart’s Law will kick in to ensure the target is hit.

In contrast, it is striking that no equivalent targets have been set for ‘harder’ test-assessed measures like the phonics check, or Year 4 times tables check, or KS2 or GCSEs.

As well as the push for phonics the last Conservative government more than doubled real terms spending on the early years between 2010 and 2024 and started a further expansion which is currently rolling out.

By September 2025, working parents will be able to claim 30 hours of government-funded childcare a week, over 38 weeks of the year, all the way through from nine months up to their child starting school.

The SEED study of early years had quite hazy and ambiguous findings - simply more childcare didn’t seem to make much difference, though higher quality early years did. But still, I can’t help thinking that more time in more formal childcare is likely to increase formal EYFS scores too - but it will be interesting to see to what extent this flows through into year one phonics.

Overall, the phonics check, as part of a push for phonics, has been exemplary policy. A strong focus on everyone mastering the basics is a key part of Asia’s school success, and we can see in the UK the strong correlation between how children do at the start and end of school. But we need to double down on it, because in many groups (and in many places) children aren’t as well prepared to learn as very similar children are in other places and groups.

Our five year old is going through all this at the moment, and there is a sort of magic as he learns to run the sounds together and begin to read. We are all so stoked when he can do it. It’s a lovely thing to be part of. We need to get that magic going for more children.

1. *I have not included Gypsy / Irish Traveller in this chart as it is a very small group, but the proportion reaching the expected standard among these groups is the lowest of all.

2. See CSJ “lost boys” project for more.

3. DFE note that: “On average across OECD countries, the average mathematics score for pupils with an immigrant background was 447 which was 30 score points lower than the average performance of pupils with a non-immigrant background (479). In England the average mathematics score for pupils with an immigrant background was 496 which is similar to the average mathematics score for pupils with a non-immigrant background (499). However, in England pupils with a second-generation immigrant background had a higher average mathematics score (509) than both pupils with a first-generation immigrant background (485) and pupils with a non-immigrant background.”