Getting more young people into education, employment and training “cannot just be about identifying the most disadvantaged young people”, ministers have been warned, as a study shows “large differences” in the paths pupils take across the country. Analysis by SchoolDash, funded by The Gatsby Foundation, demonstrates the scale of the challenge in Andy Burnham’s vow to bring down the NEET rate, with pupils in “poor urban” communities the most likely to be left behind. These pupils are also least likely to study triple science at GCSE and take on science, technology, engineering and maths (STEM) subjects at A-level. SchoolDash has published the outcomes of a research project which “clustered” schools based on deprivation indicators such as crime, health and income to create six groups of schools with similar socioeconomic characteristics. It then used a machine-learning algorithm to look at pupil performance and destinations, qualification entries, absence, exclusions and other factors. It comes as the Department for Education and Ofsted prepare to roll out their own “similar schools” model so settings can more easily be compared to those facing a similar context. NEET problem Secondary school pupils in poor urban schools – mostly in northern cities outside London – were more than twice as likely to end up not in education, employment or training (NEET) after completing key stage 4 (8.7 per cent) as their counterparts in affluent suburban schools (3.9 per cent). The study defines affluent suburban schools as those in “commuter-belt suburbs and small towns, mostly inland with long journeys to work”. Poor suburban and coastal schools also saw higher NEET rates than most other areas (7.2 per cent). The study found NEET outcomes at age 16 and 18 both vary by income deprivation. But rates of pupils going into further education “split the country in two”. They are high for suburban middle England schools, poor suburban and coastal schools and poor urban schools, but lower for the other groups. Affluent suburban schools and those with better-off intakes with high housing costs showed high rates of progression into higher education, as did urban London schools. The others had lower rates. But affluent areas with high housing costs and urban London “also show low proportions of 18-year-olds going on to employment destinations”. “Assuming that individual talent and propensities are evenly distributed, why such large differences in educational and career paths?” wrote SchoolDash founder Timo Hannay. ‘Not just about the most disadvantaged’ Jenni French, head of STEM in schools at The Gatsby Foundation, said the analysis “shows why tackling NEET cannot just be about identifying the most disadvantaged young people. “Yes, NEET rates rise with deprivation, but the routes young people take into further education, higher education and employment vary considerably between different types of places. “If government wants to reduce the number of young people who are not in education, employment or training, it needs to understand the different local barriers and opportunities young people face.” Government reforms also seek to boost take-up of STEM subjects. Ministers have told schools they will be expected to work towards offering triple science GCSE as standard. SchoolDash’s figures show the proportion of pupils entering triple science GCSE was 31 per cent in affluent suburban schools, compared to just 17.8 per cent in poor urban schools. Figures also show the proportion of pupils in different school clusters taking STEM subjects, such as the sciences, computing and maths, at A-level. In computing, urban London schools had the highest rates of entry (7.4 per cent), while poor urban schools had the lowest (4 per cent). In maths, schools serving affluent communities but with high housing costs had the highest entries (37.4 per cent), closely followed by urban London schools (36.6 per cent). Again, poor urban schools had the lowest rates (25 per cent). French said the differences in STEM participation were “particularly striking”. “The government has committed both to widening access to triple science and to expanding technical education from age 14. “If those ambitions are to translate into genuine opportunity, policymakers need to understand why access to subjects and routes varies so much between different places.” ‘Poor urban’ areas left behind Overall the analysis shows pupils in “poor urban” schools underperforming their counterparts elsewhere across most metrics. These schools have the lowest attainment and progress scores, high exclusions and absences. They also have the youngest teachers, with only 16.8 per cent over 50, as well as the highest sickness and vacancy rates and the highest agency spend. The study also tracks variations between clusters of schools in primary performance. In affluent suburban primary schools, 67.7 per cent achieve the expected standard in reading, writing and maths. In poor urban primaries the figure is 58.8 per cent. Again, absence rates were higher in poor urban primary schools (6.1 per cent) than affluent suburban (4.6 per cent). The gap in persistent absence is even larger – 17.4 per cent in poor urban schools and 9.4 per cent in affluent suburban settings. ‘National patterns’ SchoolDash’s clustering work aims to find a more useful guide to the circumstances faced by schools than pupil premium data, which measures the number of children eligible for free school meals at any point in the past six years. The pupil premium is widely used as a proxy for disadvantage, but its usefulness is limited. The government recently announced income data will replace free school meals as a trigger for disadvantage funding. And other studies have sought to look at persistent disadvantage. Hannay said: “We often treat schools with similar proportions of pupils on free school meals as if they face similar challenges, but this isn’t necessarily true: the educational effects of poverty differ between the countryside and the city, the north and the south, coastal areas and inland locations. “Yet, as we show, there are national patterns that can be teased out and used to support decision-making. As well as providing specific insights, we hope that the novel approach we have used here acts as an example of how to use AI in the service of better-informed policy.”