Assessing secondary science

18th September 2026 by Timo Hannay [link]

Update 21st September 2026: Here is an update from our collaborators at the Gatsby Charitable Foundation.

Update 18th September 2026: See also this coverage in Schools Week.

How effective is science teaching in England's secondary schools? Aside from looking at GCSE, A-level and PISA results, which tell us about ultimate academic outcomes, there is relatively little information about what's working, what isn't and how much this varies by school. Ofsted sends its inspectors to view lessons, but doesn't provide grades by subject, understandably limiting itself to qualitative remarks if and when their inspectors see fit to comment.

Yet recent developments with large language models (LLMs) mean that we can now treat such qualitative information much more like a quantitative data set. That is exactly what we will do in this post, assessing the state of science teaching through the lens of comments made by Ofsted inspectors. Our thanks to the Gatsby Charitable Foundation for commissioning and supporting this work.

In summary, we find that:

Schools serving higher proportions of poor families or lower-attaining pupils receive consistently less positive comments from Ofsted inspectors regarding their provision of science education. The gap is largest for academic attainment, which is consistent with the already well-documented trend of worse exam outcomes, but there are also clear differences for teaching quality, enrichment, assessment and curriculum, among other themes.
Negative comments about science teaching tended to focus on insufficient specialist knowledge, uninspiring lessons, low expectations, unhelpful feedback and lack of practical engagement.
Negative comments about the science curriculum mentioned inadequate planning, timetabling problems, poor resources and insufficiently flexible pathways for both the most and least able pupils.
Ofsted's observations about curriculum pathways are consistent with huge disparities in GCSE entry rates for double and triple science across geographical regions and school types. These seem to be especially influenced by deprivation levels.
The overall picture is one not only of uneven exam outcomes, which can already be seen in GCSE and A-level results, but also of disparities in assessed teaching quality and curriculum provision across different regions and school types.

For more detail, please read on.

Inspections and mentions

We processed a total of 40,641 Ofsted reports for secondary schools that had been published online in PDF form on or before 27th June 2026. These were fetched and the contents converted to plain text, which was then searched for sentences mentioning "science", "biology", "chemistry" or "physics." After assessing which reports yielded the most keyword matches, we selected eight of a total of 16 distinct inspection types for inclusion in the full analysis that follows. These comprised of 31,748 reports, broken down as follows:

Full inspection (12,820 reports, accounting for 84.7% of sentences mentioning science)
School inspection (5,475, 3.4%)
Standard inspection (3,972, 3.9%)
Monitoring visit (3,809, 2.7%)
Curriculum and development visit (2,572, 2.4%)
Additional inspection (1,645, 0.5%)
Short inspection (1,338, 1.2%)
Inspection (117, 0.7%)

The other eight inspection types, accounting for 1,741 reports, each yielded less than 0.5% of science sentences, so were left out. Also omitted were 7,131 mostly older reports whose URNs (school identifiers) had been retired by the Department for Education and could not be assigned a type, along with a further 21 reports that predated our cutoff date of September 1999.

Figure 1 shows how these different report types have waxed and waned over the years. 'Full inspections' predominated until September 2019, after which 'school inspections' took over. 'Standard inspections' are rare by comparison, but have become more frequent in recent years. 'Monitoring visits' peaked in the mid-2010s, then again during the COVID-19 pandemic. 'Curriculum and development visits' were relatively common between 2006 and 2012, disappearing altogether in late 2015. Following this, 'additional inspections' and 'short inspections' became more common, though the latter has now also disappeared. Plain old 'inspections' made an appearance in the early 2000s, but have not been seen since then. (Show all inspection types again.) For further details on the changing face of inspection types over the years, see our fuller 2023 analysis of Ofsted inspections.

Ofsted's current inspection framework took effect in November 2025. It introduced revised inspection categories and a new five-point scale, whilst eschewing overall school performance ratings (already stopped in September 2024). These changes do not directly affect the analysis presented here, which relies on the text of Ofsted's reports rather than grades assigned, though changes to the framework inevitably influence the themes that inspectors comment on.

(Click on the figure legend to turn individual inspection types on or off; double-click to show one on its own. Hover over the graph to see corresponding values.)

Figure 1: Monthly numbers of Ofsted inspections of secondary schools in England for selected inspection types (1999-2026)

Sources: Ofsted; SchoolDash analysis.

The sentences that matched our science-related keywords ("science", "biology", "chemistry" and "physics") were extracted and manually analysed with the help of an LLM (gemini-3.6-flash) to create a taxonomy of the 10 themes most frequently mentioned by Ofsted inspectors in relation to science education:

Attainment: Academic attainment and examination outcomes
Teaching: Quality and efficacy of classroom teaching
Staffing: Teacher recruitment and staffing stability
Practicals: Practical science investigation and enquiry skills
Curriculum: Curriculum architecture and qualification pathways
Assessment: Assessment, marking and feedback tracking
Laboratories: Specialised laboratory infrastructure and resources
Disciplinary literacy: Disciplinary literacy and scientific vocabulary
Enrichment: Extracurricular enrichment and STEM outreach
Safety: Laboratory safety and operational risk mitigation

We then fed the full text of each Ofsted report to another LLM (gpt-5.4-mini) with instructions to determine which of the above themes were mentioned in the context of science education. For those that were, the LLM also assessed whether the overall sentiment was positive, neutral or negative. Note that, unlike the taxonomy creation step described above, this did not rely on simple keyword matching, but on holistic textual analysis of each Ofsted report by the LLM. A random subset of responses was manually inspected and found to be of comparable quality to that from a diligent human curator; no egregious errors were evident.

Figure 2 shows the frequency with which each theme was mentioned across all Ofsted reports analysed, and how this has changed over time. Teaching, staffing and curriculum are perennial themes in science education that have all been relatively frequently mentioned over the years (give or take a dip during the pandemic). Conversely, practicals, laboratories, disciplinary literacy and safety were popular until the mid-2000s, but are no longer mentioned very much. Assessment is somewhere in the middle, having suffered a decline, but not a precipitous one. Finally, attainment and enrichment have both fallen, but also seen a recent uptick under the current Ofsted inspection framework (ie, since November 2025).

(Use the menu below to switch between themes. Hover over the graph to see corresponding values.)

Figure 2: Monthly proportions of Ofsted reports mentioning science education by theme (1999-2026)

Notes: Includes mentions of science, biology, chemistry and/or physics. Months with fewer than 10 Ofsted reports have been omitted.

Sources: Ofsted; SchoolDash analysis.

Do the frequencies with which these science education themes are mentioned vary by school type or location? In a word, sometimes.

Figure 3 shows the frequencies with which each theme was mentioned across a range of different school characteristics. Note that in this case we are looking only at the most recent Ofsted reports for schools that are currently open. So this is not a deep historical archive, it is currently applicable data comprising of a total of 5,240 Ofsted reports.

Perhaps the most of obvious trend to examine is by overall Ofsted rating (for those schools last inspected before September 2025). Inspectors of schools with low ratings are more likely to mention staffing and assessment, but less likely to refer to enrichment. Another salient axis is the in-school deprivation level. Inspections of schools serving high-deprivation communities are less likely to mention any of these themes, from attainment to safety. (This may be in part because, as we shall see below, the average inspection report tends to dwell more on the positive than the negative.) Predictably, trends by prior attainment and Key Stage 4 (KS4) attainment show broadly similar trends to those for in-school deprivation. For example, enrichment is mentioned more frequently for schools with high prior or KS4 attainment. Small and rural schools also tend to get fewer comments across most of these themes – see for example the trends for attainment; the exceptions are laboratories and safety.

(Use the menus below to switch between themes and school types. Hover over the columns to see data values.)

Figure 3: Prevalence of Ofsted themes by secondary school type

Show: by

Notes: Most recent Ofsted reports for currently open secondary schools. Sample size: 5,240 (including 3,254 mainstream state schools, 819 state special schools and 1,167 Ofsted-inspected independent schools). School deprivation figures based on pupils' eligibility for free school meals, with bands defined by the DfE: low means less than 20%, high means more than 35%. Local deprivation figures based on the mean IDACI of postcodes within a 4km radius of each school, with schools then divided into three roughly equally sized groups. A small proportion of low-attainers means less than 12% and a high proportion means more than 18%, as reported by the DfE. Low Key Stage 4 attainment means an Attainment 8 score of less than 43.1 and high attainment means a score of more than 49.1. A low proportion of ethnic-minority pupils means 10% or less, while a high proportion means more than 50%. A low proportion of EAL pupils means less than 4% and a high proportion means more than 15%. Small schools have fewer than 700 pupils, large ones have more than 1,200. Small MATs are those with 10 or fewer schools. Urban, suburban and rural groups use ONS rural-urban categories applied to school postcodes. Coastal schools are those located 5km or less from the shoreline.

Sources: Ofsted; SchoolDash analysis.

All sentimental

So far we have examined how frequently Ofsted reports mention common themes in science education, but not what they have said about it. This section will look at the sentiment (positive, neutral or negative valance) of Ofsted inspectors' comments.

Figure 4 shows the proportions of all mentions (including neutral ones) that were positive (blue line), negative (red), and the percentage-point difference between them (black). (Show all lines again.) In general, there have been more positive comments than negative ones. Staffing, laboratories and safety have tended to show more positive comments in recent years, while practicals have shown fewer – though bear in mind that the last three themes in that list have also become less likely to get mentioned at all. The sentiments associated with other themes – attainment, teaching, curriculum, assessment, disciplinary literacy and enrichment – have generally stayed flat over the years, though they all display considerable month-to-month fluctuations and some discontinuities, such as during the pandemic.

(Use the menu below to switch between themes. Click on the figure legend to show or hide each line; double-click to show a line on its own. Hover over the graph to see corresponding values.)

Figure 4: Monthly proportions of Ofsted reports mentioning science by theme (1999-2026)

Notes: Includes mentions of science, biology, chemistry and/or physics. Months with fewer than 10 Ofsted reports have been omitted.

Sources: Ofsted; SchoolDash analysis.

We can also examine how these sentiments vary by school type or location. Figure 5 shows the net sentiment (percentage of positive comments minus percentage of negative comments) varies across schools with different characteristics. As in Figure 3 it includes only the most recent Ofsted report for schools that are currently open, so reflects currently applicable data.

Unsurprisingly, there are big differences in sentiment by overall Ofsted rating (where available). These are particularly large for attainment, teaching, staffing, curriculum, assessment and enrichment, suggesting that these may have been among the bigger drivers of the ratings ultimately awarded.

It's perhaps more informative to look at the differences by in-school deprivation (and, relatedly, for prior and KS4 attainment). Here there are considerable disparities in the sentiments expressed about attainment (to be expected if exam results are genuinely worse) but also teaching, staffing and enrichment. The implication is that Ofsted judged those aspects of science education to be of lower quality in schools serving poorer, lower-attaining pupils than across the secondary school system as a whole.

Small and rural schools show similar trends in sentiments expressed about science teaching and staffing trends, as do schools with lower proportions of ethnic-minority and EAL pupils. Interestingly, Ofsted inspectors seem quite positive about the science enrichment activities at rural schools, though not at small schools more generally. This may speak to opportunities afforded by the local environment where they cannot be provided by the school itself.

(Use the menus below to switch between themes and school types. Hover over the columns to see data values.)

Figure 5: Sentiment of Ofsted themes by secondary school type

Show: by

Note: See notes to Figure 3.

Sources: Ofsted; SchoolDash analysis.

What makes a good science teacher?

In order to better understand the nature of these various positive and negative comments about science education, it is useful to review their contents. To make this manageable, we will limit ourselves to comments about (a) science teaching and (b) the science curriculum. Among the 10,229 positive inspector comments that were categorised under science "teaching", the most frequent topics were:

Strong subject knowledge and specialist expertise: Deep subject expertise, clear explanations of complex theories and infectious enthusiasm.
High-quality teaching and student progress: Strong, effective teaching, driving high levels of pupil progress, strong conceptual understanding and above-average examination results.
Engaging practical work and investigations: Hands-on experiments, field investigations, demonstrations and real-world applications that brought abstract concepts to life and sparked student curiosity.
Effective questioning and misconception management: Probing, targeted and challenging questioning techniques to test understanding, address misconceptions and promote higher-order scientific reasoning.
Integration of cross-curricular skills: Successfully incorporating scientific literacy (mastering technical vocabulary and writing), numeracy (applying formulas, calculations and graph interpretation) and digital tools (interactive whiteboards, sensors and data loggers).

Conversely, the most frequent topics about science teaching among the 3,288 negative comments were:

Staffing instability and lack of specialists: Recruitment challenges, high staff turnover and a reliance on temporary, supply or non-specialist teachers, disrupting learning continuity and leaving gaps in student knowledge.
Inconsistent teaching quality and low expectations: Variable quality of instruction across key stages, departments and individual teachers, with uninspiring lessons that lacked ambition or failed to stretch higher-attaining pupils.
Passive learning and uninspiring methods: Lessons relied too heavily on teacher-led lecturing, textbooks and passive tasks, such as copying notes, completing basic worksheets or cutting-and-pasting.
Weak assessment, marking and feedback: Teachers failed to use assessment effectively to gauge prior learning, adapt future tasks or correct misconceptions, with irregular or unhelpful marking that lacked actionable advice.
Lack of practical enquiry and engagement: Failure to provide sufficient opportunities for independent investigation, data analysis or practical experiments, leading to student boredom, off-task behaviour and untidy or incomplete work.

Curriculum comments

The most common topics in the 12,034 positive inspector comments about the science "curriculum" were:

Impact of specialist status and funding: Designated specialist status (such as science, maths, technology or STEM college status) enhanced resources, improved standards and drove curriculum innovation.
Broad and flexible curriculum pathways: Offering tailored academic and vocational routes at Key Stage 4 and post-16. Options for BTEC applied science, GNVQs, core/additional science and vocational pathways alongside traditional GCSEs.
Provision for high attainers and triple science: Stretching academically able students by providing early entry examinations or offering three separate single sciences (biology, chemistry and physics).
Effective curriculum sequencing and continuity: Well-planned, modular or sequenced curricula, allowing pupils to build directly on prior knowledge across key stages. Clear transition planning from primary feeder schools.
Enrichment activities and external links: Out-of-classroom learning, such as lunchtime or after-school science clubs, university partnerships, STEM projects, field trips and guest speakers.

And the topics most commonly mentioned in the 2,344 negative comments about curriculum were:

Inadequate curriculum planning and sequencing: Poorly planned, fragmented or uncoordinated schemes of work. Lack of logical progression in teaching concepts, failure to build systematically on prior learning and long gaps between topic coverage, especially when rotating between biology, chemistry and physics.
Insufficient teaching time and timetabling issues: Time allocated to science lessons fell below national recommendations. Timetabling flaws – such as cramming multiple lessons into single days or split classes across multiple teachers – restricted coverage of the full National Curriculum and affected pupil progress.
Curriculum narrowing, inappropriate pathways and early entry: Poor course choices that limited pupil potential or career routes. Lower-attaining or SEN pupils were restricted to inappropriate vocational options, while higher-attaining pupils were denied access to triple science (separate biology, chemistry and physics exams). Forcing all pupils onto academic or triple science routes regardless of capability was also criticised.
Underdeveloped practical work, investigation and ICT skills: Lack of practical experiments, enquiry skills and scientific investigation in lessons. Failure to meet statutory requirements for integrating ICT into science, such as data logging and sensor technology.
Resource shortages, inadequate accommodation and staffing issues: Lack of specialist facilities, such as cramped or missing laboratories, inadequate basic equipment, shortages of subject-specialist teachers (particularly in physics), and weak subject leadership or lack of technician support.

Looking at both teaching and curriculum together, the most salient topics separating the best schools from the rest appear to be teacher expertise, enthusiasm and imagination, along with well-designed, flexible curricula that cut across individual subjects and provide suitable practical and enrichment activities. Ability-appropriate curriculum options were also deemed important.

Double or triple?

Since the provision of double and triple science at GCSE is clearly a salient issue for Ofsted, it is worth reviewing how this varies greatly across geographical locations and school types.

Figure 6 show the mean entry rates for GCSE biology, chemistry, physics and double science by school type. There are notable regional disparities, with triple science (see biology, chemistry and physics) showing considerably higher entry rates in southern parts of the country, while double science predominates in the north and midlands. But the differences by school deprivation level are even more stark (see biology, chemistry and physics), with much higher entry rates in schools serving more affluent populations. double science shows a more nuanced picture, with a peak amongst schools serving medium-deprivation communities – a consequence of the fact that many pupils in the highest-deprivation schools don't sit any science GCSEs at all.

There are other interesting patterns by local deprivation, prior attainment, Key Stage 4 attainment, Ofsted rating, proportion of ethnic minority pupils, proportion of EAL pupils, school size, multi-academy trust (MAT) status, academy type, grammar schools, presence or absence of a sixth form, faith status, urban status and distance to the coast – but we will leave those for you to explore. 🕵️

(Use the menus below to switch between subjects and school types. Hover over the columns to see data values.)

Figure 6: GCSE science entry rates by secondary school type

Show: by

Note: See notes to Figure 3.

Sources: Ofsted; SchoolDash analysis.

The overall picture is one of uneven provision of science education, at least as seen through the eyes of Ofsted inspectors, with particualrly notable differences in teaching quality and curriculum flexibility. We have long been able to see the consequences in GCSE and A-level results, but by analysing Ofsted reports we can get a glimpse upstream at factors that may be contributing to those outcomes. Of course, as described in our recent analysis of Ofsted inspection grades under the latest framework, it is possible that schools serving poorer communities are not performing any worse, but rather being judged more harshly. The available data cannot distinguish between those two possibilities, so we leave it to you, our readers, to decide – noting only that either would be a serious cause for concern.

We hope that this post has provided not just a window into national disparities in science education, but also a glimpse into the potential for turning qualitative sources, such as the huge archive of Ofsted reports, into semi-quantitative data sets amenable to deeper research. As always, we welcome your thoughts: [email protected].

Does the new Ofsted framework take adequate account of pupil characteristics?

11th September 2026 by Timo Hannay [link]

Update 11th September 2026: See also this coverage from Tes.

Since last November, Ofsted has employed a new inspection framework. The concept of overall grades had already been banished more than a year earlier, but now the old four-point scale (running from Outstanding to Inadequate) was replaced with a five-point scale (from Exceptional to Urgent improvement) and half a dozen or more newly defined evaluation areas.

Importantly, inspections were also to be "grounded in a clear understanding of each provider's unique circumstances," with inspectors to "reflect on a provider's context and priorities." Is this working? The following analysis, based on more than 2,000 inspection reports published so far, suggests perhaps not.

Grade distributions

First, let's look at the way that grades have been apportioned across all schools. These are shown in Figure 1.

It is immediately clear that the new 'Exceptional' grade (leftmost column) is vanishingly rare: around 1% of schools so far, compared to more like 5-10% for the previous 'Outstanding' grade – though that proportion had tended to fall in more recent years (see our previous analysis). The new 'Strong standard' straddles the old 'Outstanding' and 'Good' grades, accounting for around 15-35% of schools, depending on the inspection area. 'Expected standard' makes up about 55-65% of schools, which roughly corresponds to the rest of the schools that would have been 'Good' in old money. 'Needs attention' (≈'Requires improvement') and 'Urgent improvement' (≈'Inadequate') more or less map onto the two lowest grades under the previous framework.

Note also that there is some variation by phase: see primary, secondary and all schools together. And there are even bigger differences by evaluation area. See: 'Leadership and governance', 'Curriculum and teaching', 'Achievement', 'Attendance and behaviour', 'Personal development and wellbeing' and 'Inclusion'.

(Use the menus below to view different school phases and evaluation areas. Hover over the columns to see corresponding data values.)

Figure 1: Distribution of Ofsted grades by phase and inspection area.

Notes: Reports published on or before 5th August 2026. Includes 36 all-through schools that are included in both primary and secondary counts.

Sources: Ofsted; SchoolDash Insights; SchoolDash analysis.

Disadvantaged?

How, if at all, does all this relate to the pupil composition of each school?

Figure 2 shows how the mean percentage of pupils eligible for free school meals (FSM) varies by Ofsted grade. This reveals somewhat different patterns by phase. Primary schools sometimes show very high FSM levels among the small numbers of 'Exceptional' schools, especially in 'Leadership', 'Curriculum' and 'Inclusion'. This makes a certain sense if schools serving poorer communities have more opportunities to demonstrate genuine excellence in those areas. Nevertheless, it's interesting to observe that for other grades the proportions of FSM pupils tends to rise as the grade gets worse. Furthermore, the trend for 'Achievement' and 'Attendance' don't show such a huge FSM spike among 'Exceptional' schools. Again, this could be justified given that academic performance and absences really do tend to be worse among schools with greater numbers of disadvantaged pupils – though it still makes you wonder how much Ofsted is really taking school context into account. ('Development' doesn't show much of a pattern at all, which is arguably what we would hope to see for all areas if school context was being fully reflected in the judgements.)

The picture for secondary schools is less ambiguous: across all inspection areas, schools with worse inspection grades tend to have higher proportions of FSM-eligible pupils (though most show a slight FSM bump for 'Exceptional' schools). This is arguably a bit less understandable, especially for 'Leadership': are we to to conclude that secondary schools with higher proportions of poor pupils are systematically less well led, or simply that Ofsted rates them more harshly? (Show all schools again.)

(Use the menus below to view different school phases and evaluation areas. Hover over the columns to see corresponding data values.)

Figure 2: Mean FSM level by Ofsted grade

Notes: Reports published on or before 5th August 2026. Includes 36 all-through schools that are included in both primary and secondary counts.

Sources: Ofsted; SchoolDash Insights; SchoolDash analysis.

Special cases

Figure 3 shows the average proportions of pupils receiving support for special educational needs (SEN), once again segmented by Ofsted grade. The trends here are clearer still: across both phases (see primary and secondary) and all inspection areas (see 'Leadership and governance', 'Curriculum and teaching', 'Achievement', 'Attendance and behaviour', 'Personal development and wellbeing' and 'Inclusion'), schools with lower Ofsted ratings almost invariably have higher incidences of SEN. Why should this be? (Show all schools again.)

(Use the menus below to view different school phases and evaluation areas. Hover over the columns to see corresponding data values.)

Figure 3: Mean SEN level by Ofsted grade

Notes: Reports published on or before 5th August 2026. Includes 36 all-through schools that are included in both primary and secondary counts.

Sources: Ofsted; SchoolDash Insights; SchoolDash analysis.

Finally, Figure 4 shows that similar patterns hold true for pupils with Education, Health and Care (EHC) plans, who constitute the relatively small (but growing) proportion assessed to have particularly serious educational needs. It might seem inevitable that this will show the same patterns as SEN in Figure 3, but that's not necessarily the case because the SEN and EHC pupil categories don't exactly co-correlate in schools. For example schools in more affluent areas tend to show lower levels of SEN but higher proportions of pupils on EHC plans (sometimes attributed, rightly or wrongly, to sharp-elbowed middle-class parents being better able to secure the necessary diagnoses). Yet sure enough, across primary and secondary schools, and most inspection areas, the worse the grade the higher the proportions of EHC plan pupils. Once again, this is perhaps hardest to justify for 'Leadership', where the trend is particularly stark for primary schools. (Show all schools again.)

(Use the menus below to view different school phases and evaluation areas. Hover over the columns to see corresponding data values.)

Figure 4: Mean EHC level by Ofsted grade

Notes: Reports published on or before 5th August 2026. Includes 36 all-through schools that are included in both primary and secondary counts.

Sources: Ofsted; SchoolDash Insights; SchoolDash analysis.

So, relatively speaking, schools that receive low Ofsted grades are more likely to have high proportions of SEN and EHC plan pupils (and, especially in the case of secondary schools, high FSM levels). It is important to bear in mind that the sample sizes for very high- and low-graded schools are currently small, so the effects seen at the extremes should, for now, be taken with a pinch of salt. Nevertheless, the overall patterns are thought-provoking, especially when it comes to Ofsted's assessment of school leadership amidst claims of greater attention to context. Is it that schools with higher SEN levels are genuinely more likely to be badly run, or simply that Ofsted is not taking adequate account of pupil characteristics when judging school leaders? It could even be a bit of both. Unfortunately we can't tell from the available data, so resolving this important distinction is left as an exercise for our ever-thoughtful readers. We'd love to hear your thoughts: [email protected].

2026