While ‘carrot and stick’ initiatives can make a difference to improving attendance, what’s really needed for long-term improvement is a more strategic approach, using data to truly understand what’s going wrong and why.
The problem is there’s so much attendance data – one year group over one academic year in our school alone takes up 300,000 rows in Excel.
But what if you could use artificial intelligence to spot trends, identify areas for improvement and perhaps even stop absenteeism before it occurs? We were already using Microsoft Copilot for exam results analysis, so we decided to give it a go.
It took us a while to get the correct prompts, simplify the data, and present it in a language parents would appreciate.
After some trial and error, we ended up with a succinct report for each pupil, an ‘attendance profile’ showing their strengths and areas for development.
Most importantly, we had a level of detail and insight into attendance that we never had before. We sent these reports to parents for the first time this term.
‘Patterns we never would have noticed’
The AI revealed patterns and trends we never would have noticed otherwise, without poring over registers manually for hours.
For example: spotting when families take children on holidays at the same time every year. Isolating where families have children absent regularly on the Fridays prior to holidays or the Mondays after.
Seeing patterns of absence across terms or individual months, and noting patterns of absences related to specific age groups or gender, or pupils from one area of the city compared to others.
Absences might also correlate with wider events, from things happening in the community, to online activity, such as the launch of a new game or a major concert.
When we know an event is likely to affect a particular group, we can contact parents in advance to emphasise the importance of attendance. We can even cross-reference weather patterns to understand how they influence attendance.
Predicting issues
AI helps us use our existing data more intelligently to predict issues before they become problems.
For example, if some families consistently take holidays at the same time each year, we can write to them months in advance to remind them not to book trips during term time.
As a result of this, we’ve adopted a raising standards leader structure to attendance, treating it with the same rigour as we do academic outcomes and pupil progress.
One person has overall responsibility for attendance data and liaises with all those involved so the message is clear and consistent.
It also means we can provide heads of year and other pastoral staff with key intelligence, so when they meet parents, they can always discuss attendance with reference to specific situations, both future and historical.
This is just the beginning of what could be a very exciting development in tackling truancy and absenteeism.
I took this approach to a meeting of Cardiff’s secondary headteachers recently, and none of them had ever heard of an approach like this. As a result, we’re going to be training their staff how to do this in their schools.
Identifying trends
And this brings up another prospect – can we work together and use AI to identify attendance trends across schools in particular areas of the city? Are there times of the month or year when there are particular issues, for example?
We’re soon going to be moving to Power BI, another Microsoft product which can take bigger data sets, which will allow us to analyse multiple years of data.
What if we could extend this approach to our feeder primaries? If we can do this kind of long-term data analysis with year 4/5/6 pupils, we may be able to solve attendance problems before they come to us in year 7.
We’re only just beginning to understand the potential of AI in education and explore its uses, but if it can have an impact on some of the more stubborn and intractable issues we face, like low attendance, then it’s definitely worth exploring further.
For example, I read with interest the announcement earlier this month that all schools in England are to be given an AI-generated target for minimum pupil attendance.
This plan also involves schools working together to learn from best practice, which is always a good idea.
But schools have so much attendance data, and if they can analyse that using AI to predict and pre-empt issues, they can have even more of an impact.
Difficult problems require innovative solutions, and we believe that’s what we have developed with our approach.
Your thoughts