10th January 2023

Using data on previous absence to predict current absence

Last term I published Education history and attendance, which found that previous absence was a strong predictor of current absence. Among pupils who had previously been persistently absent, the rate of persistent absence in autumn and spring of academic year 2021-2022 was 51%, over three times higher than the rate of persistent absence among pupils who hadn’t been persistently absent (14%).

In this paper I outlined my ambition for making better use of attendance data to effectively target support to pupils with low attendance. A rapid evidence assessment from Education Endowment Foundation on attendance interventions found that targeted parental engagement strategies were more effective than untargeted parental engagement strategies. This review also found that responsive interventions, where a member of staff targets their approach specifically to the needs of an individual pupil, were effective. More needs to be done to target limited attendance resource, providing support or sign-posting existing support, to pupils likely to become persistently absent.

Making use of data needs to go beyond utilising the current terms attendance data from day one of a new term. Schools must have access to children’s previous attendance history so that they can plan proactively, even before the start of a new term. Where schools don’t have access to historical data, for example when a child has moved schools, the local authority should work with them collaboratively to share data resources and join up the dots. Schools should then use this evidence, alongside conversations with their families and knowledge of their local community to implement the right plan of support for every child.

Today we’re publishing code that can be used by schools and Multi-Academy Trusts to better understand patterns in their pupils attendance data. This code is in R, a statistical software, but similar analysis can be done in excel or using other software.