Are children really going backwards between two and five?

Nathan Hodson · September 2026

Around 85% of two-year-olds are at expected level on the ASQ. Around 68% of five-year-olds reach GLD. The gap between those numbers is the longest unobserved stretch in a child’s early years — and councils are responding in three different ways.

Almost every plan I read reports two figures side by side: the share of children at or above expected level in all five domains of the Ages and Stages Questionnaire at their 2–2½ year review, and the share reaching a good level of development at the end of Reception. In most places the first is in the eighties and the second in the sixties. Wirral says it directly: “The drop from 83.3% reaching their milestones at age 2 to 66% at GLD highlights widening gaps as children near school age.” Rotherham (88.2% at two; 64.3% at five): “Developmental inequalities persist and widen between ages 2½ and school entry.” Manchester is above the national average at two and yet “this data is not transferring into GLD outcomes”. Cheshire West and Chester: “this strength isn’t translating into GLD at age 5… a drop-off between ages 3–5.” Westmorland and Furness falls from 88.8% to 67.4%; Bristol from 88% to 68.7%; Luton from 82.7% to 64%.

Figure. Share at expected level at 2–2½ (blue) and GLD at five (orange), for every plan giving both. Nottingham is the exception.

There are three explanations in the plans, and each leads to a different response.

But first lets look closely at the pattern. Across Local Authorities there is much higher variation at the two-year-olds than five-year-olds, such that those with lowest scores among two-year-olds actually see increases by five. In Nottingham we see an increase: 97.1% of its two-year-olds received the ASQ, but only 62.5% were at expected level in all five areas, against a national 81.4% — and its GLD at 5 years old is 65.1%, actually higher than among two-year-olds. Telford and Wrekin (64.3% at two, 67.3% at five) and Shropshire (65.6%, 68%) show the same inversion: the two-year figure is the lower one. Overall, the standard deviation at the two-year-old test is 6.7 percentage points and the standard deviation at five years old is 2.8 percentage points. There has to be some explanation for this strange conversion.

Explanation one: the two-year old measures are not working

It could be that measures at two-years-old do not measure the same thing as those at five-years-old. This is an alternative to true, massive, differences among two-year-olds which all even out by five.

Some Local Authorities have mitigated this by improving and integrating assessment at two-years-old. Sandwell and Kent combine the ASQ with the early years progress check so that health visitor and key person assess together —Tower Hamlets is reintroducing integrated reviews in settings from 2027. Bromley’s deep dive into a hundred cases led to a similar integrated two-year review. Westminster and Kensington & Chelsea change the language around these reviews to try to improve the quality of assessments, replacing jargon with “life skills” language, because “many parents interpret ‘school readiness’ as academic (phonics/reading) rather than independence and self-regulation”. Improvements in measurement policies could lead to more consistent (and plausible) measures.

Explanation two: selection bias at two-years-old

Coverage is a major problem across Local Authorities. Kirklees’s ASQ result looks fine at 82% — but completion is 69.4%, “24.5 points below England”. Worcestershire: only 51.8% of children received the check by 30 months. Cumberland: attendance rose from 48.4% to 75% in a year. Kingston: 29% non-attendance. North Northamptonshire records that DHSC itself flagged “Low coverage of 2 to 2½ year reviews using ASQ-3”. A council with 80% coverage and 85% at expected level only really knows 68% of its two-year-olds have reached the expected level unless they can show that non-attendance is uncorrelated with development level.

Figure. Coverage of the 2–2½ review as each plan reports it. Orange: below 80%.

Practical behavioural responses could reduce this bias. Kirklees hands out toothbrushing packs at the review, “helping to encourage attendance”. Lincolnshire lets parents self-book. Hillingdon and Haringey set numeric coverage targets. Brent sends termly lists of children who missed the review to the local authority and phones the families. These might serve to reduce non-engagement, reducing the risk of non-random omissions.

Explanation three: something really happens between two and five

Local Authorities such as Rotherham are concerned about a true change with an impact on equality: “Developmental inequalities persist and widen between ages 2½ and school entry." If the slide is real, one response is another review to identify opportunities to intervene and a cluster of councils has added one. Barking and Dagenham is reinstating a 3–3½ year ASQ review “previously cut when funding reduced”. Trafford adds a health visitor check at three and four in high-need areas. Bath and North East Somerset adds a contact at three where needs are identified. Sefton adds a three-year check. Oxfordshire adds “a new universal 4 year review”. Hertfordshire will “start a new ‘School Readiness’ check in the Health Visiting Service”. Manchester is exploring “a school-readiness check at age 3”. Herefordshire already reports “% Health Visitor visits completed at 3 years (new)” with targets of 85, 90 and 95%, and its family hubs plan names a “new 3–4 years development check”. Blackpool has an existing 3–3½ check; Bury adds a targeted 3–3½ health visitor contact for children with SEND; Calderdale’s health visitors will “offer an additional early language screening using the WellComm tool” at three for children not in a setting; Southampton commits to “targeted developmental reviews beyond age two”; Torbay builds a four-year stage into its pathway.

Figure. The mandated Healthy Child Programme reviews (black) and the extra contacts councils have written into their plans (blue).

Given the strange finding that development levels more widely across Local Authorities at two-years-old than at five-years old, it is valuable to add an additional monitoring point in the middle to make more sense of child’s journey.

Three Years to Produce Change

As Local Authorities set out to hit their targets for Good Level of Development, strategies have been built around identifying the 300 or so children at risk of missing out. If the data at two-years-old is low quality, incomplete or distorted by an intervening event, then targeted strategies will be significantly undermined. In this case, further measurement in between will be key, along with universal approaches which can be scaled fast, such as digital tools like Easy Peasy, 50ThingsToDoBeforeYou’re5, or our app Pause.