Ofqual has refused to state whether an algorithm is being used to flag up schools with unusual GCSE and A-level results this year for quality assurance purposes.

Tes understands - after speaking with sources close to the discussion - that a combination of data processing and human oversight is being used to highlight schools whose results fall outside of a particular grading “threshold”.

GCSEs: ‘Schools have been well and truly dumped on’

News: ‘Obviously wrong’ A-level grades ‘allowed’ by Ofqual

Exclusive: 94 per cent of teachers grading GCSEs report problems

But when Ofqual was asked to confirm whether an algorithm was being used to flag up schools for further investigation, it refused to comment and directed Tes to its published information for headteachers, as well as guidance for the 2021 exams issued by the Joint Council for Qualifications.

However, experts have said there is a lack of transparency over the quality assurance processes for GCSE and A level results in 2021.

GCSEs and A levels 2021: Ofqual ‘should be more transparent’ over grade checks

In Ofqual’s guide for headteachers, it says that: “Where the overall results at GCSE or A level look very different from recent years (2017, 2018 or 2019), centres should record the likely reasons for this, as exam boards might ask to see this if the centre is selected for external quality assurance.

“Exam boards will target their quality assurance based on a number of factors, including where a centre’s results are considerably lower or higher than recent years.”

It adds: “Exam boards will prioritise for quality assurance checks those centres where results are more out of line with their historical results than other centres nationally, including where grades are lower.”

Dave Thomson, a chief statistician at FFT Datalab, said that there needed to be more transparency over the quality assurance processes.

“The JCQ guidance referred to [schools] checking results against averages from 2017 to 2019. They didn’t really give any indication of what they wanted schools to do other than look at how averages across all subjects compared and didn’t give guidance on what to do if they were higher or lower or what they were then going to do with this,” he said.

Mr Thomson said in a blogpost in May that he had found weak correlation between the 2017 and 2019 results with a lot of variability for small subjects.

He said this variability could be an issue if schools were flagged using this data in 2021.

But he added: “Last year, with the algorithm, that didn’t work very well, but this year if results are flagged and you’ve got a human being looking into it and looking at the work produced and looking at the school’s policy, that might be alright.

“The fact that schools are being flagged might not be a problem but it’s how then the decisions are made about whether to accept the results or modify them, and I don’t think we know very much about that process.”

He added that in terms of how grades might be modified, “it’s certainly something they ought to be more transparent about”.

Ofqual has been contacted for comment.