England/08 august 2020/By Richard Adams Education editor/Source: https://www.theguardian.com/
Exclusive: Assessments likely to be adjusted down before students receive results – analysis.
Nearly 40% of A-level grades submitted by teachers are set to be downgraded when exam results in England are published next week, the Guardian has learned, as criticism intensifies of this year’s makeshift results.
Analysis of the algorithm and data used by the exam regulator Ofqual to distribute grades after the cancellation of exams amid the coronavirus pandemic found that a net 39% of assessments of A-level grades by teachers are likely to be adjusted down before students receive their results.
That would mean nearly 300,000 A-levels issued are lower than the teacher assessment of the more than 730,000 A-level entries in England this summer.
Including GCSEs, which are expected to have a similar downgrade rate, close to a net 2m teacher assessments will be adjusted downwards and in many cases ignored completely.
There was uproar in Scotland this week when the exams authority rejected nearly 124,000 grade recommendations from teachers – a quarter of the total – but unlike in Scotland, English pupils are barred from appealing against their results on academic grounds.
Grades will instead be issued according to Ofqual’s statistical model, relying on a school’s recent exam history and each pupil’s previous exam results, to replace the exams scrapped by the government after schools were closed because of the coronavirus lockdown.
Those most at risk of receiving revised grades appear to be students on the border between B and C grades, and between C and D grades, and pupils at comprehensive schools with wide variations in attainment or patchy outcomes in courses over the three previous years of data that Ofqual is using to cap individual school results.
Teachers will still have a significant influence on how grades are distributed in each school, having compiled the rankings that will determine which pupils receive the final grades allocated by Ofqual for their course.
Headteachers and exam officials in England say they fear a storm of controversy even worse than that which has engulfed Scotland, where a quarter of teacher predictions were adjusted by the Scottish Qualifications Authority.
Experts say that as Ofqual has barred individual pupils from appealing against their grades on academic grounds, families should not waste time complaining but instead contact college or university admissions offices to confirm their places in the event of unexpectedly poor grades.
Tim Oates, group director of research and development at the exam board Cambridge Assessment, said: “Grades have been awarded this year by combining lots of data, including the rank order and the grades submitted by teachers. We have seen from Scotland’s press coverage that it’s all too easy to fixate on the difference between the teacher-assessed grades and the final grades. But it’s a misleading distraction and misinforms the public. The teacher grades were an important part of the process but always only going to be a part.
“On results day, energy should be channelled into how each young person can progress swiftly with what they have been awarded, rather than time lost on agonising over an apparently controversial but fundamentally misleading difference between teacher grades and final grades.”
Statisticians have criticised Ofqual’s algorithm, saying it does not have sufficient data to award grades fairly to most state schools in England, because of wide variations in results within schools and between years.
The Royal Statistical Society has called for an urgent review of the statistical procedures used in England and Scotland, to be carried out by the UK Statistics Authority. “This should consider both the substantive issues of the data used and the adjustment algorithms of the various nations, but also whether greater transparency would have been possible and beneficial,” the society said.
Huy Duong, the parent of an A-level candidate and a former medical statistician, said he has analysed Ofqual’s published data and comments to calculate that 39% of grades between A* and D will be lower than the teacher assessments. Duong’s findings were privately confirmed to the Guardian by exam officials.
Duong’s analysis is based on Ofqual’s statement that A-level grades overall will improve by 2% this summer, and that the submitted teacher assessments, known as centre-assessed grades, would have resulted in 12% inflation in higher grades.
“It gives the public the impression that in most cases the grades the student receive would still be the predicted grades. However, closer analysis shows that this is not true,” Duong said.
In response, a spokesperson for Ofqual said: “From the data that we have reviewed we expect the majority of grades students receive will be the same as their centre assessment grades, and almost all grades students receive will be the same as the centre assessment grades or within one grade. The exact proportions vary by subject and we will publish the figures on results day.”
But Duong also found that a comprehensive secondary school can have a huge annual variation in results for individual courses, with the small numbers of entries involved suggesting that Ofqual’s decisions are statistically invalid.
For instance at Matthew Arnold school in west Oxford, a comprehensive academy, the proportion of A* grades achieved by pupils in a popular subject such as English literature varied from one in 19 to three in 10, or from 5% to 30%, between 2017 and 2019.
Duong said: “These fluctuations mean that Ofqual’s statistical modelling cannot make sense. The problem is that any statistical model is only as good as the data you feed it, and for a typical comprehensive school, there simply isn’t enough A-level data from 2017 to 2019 for any statistical modelling.”
In response, Ofqual said: “For A-level, three years of historical results inform the standardisation of grades. You can think of this as an averaging across the years of data.”
Uncertainty over the grades being awarded has led universities to say they will relax offers to prospective students and make use of other data.
Admissions officers for Peterhouse College, Cambridge, said during a Q&A on the Student Room website: “We are looking into which schools (from among our offer holders) are most likely to be adversely affected by the system of awarded grades so we are in a better position to make nuanced decisions when we get the results.”
In a significant sign that it recognised the controversy likely to erupt on Thursday when students receive A-level results, Ofqual this week changed its position and said schools would be able to appeal if they expected “a very different pattern of grades to results in previous years”. But Ofqual will not allow individual students to appeal against their grades on academic grounds, as they can in Scotland.
For A-level pupils, teacher assessments will only be used to help set grades on small courses, with five or fewer candidates. On larger courses, teacher assessment will play little or no role, with grades being awarded instead based on a school’s recent exam history and each pupil’s previous exam results.
The pupils most likely to benefit from teacher assessments will be those taking courses with very small entries of five pupils or less, such as German or music. Those taking popular subjects such as maths or biology, with more than 15 pupils, will be exposed to Ofqual’s algorithm and the teacher assessment will be ignored.
Kate Green, Labour’s shadow education secretary, said: “The government should have ensured that Ofqual had a robust appeals system in place from the beginning, instead of announcing one a matter of days before A-level results. They must do far more to ensure the system is genuinely fair.”
Students unhappy with their results have the option of sitting exams in autumn. But with students entering university or college, few are expected to do so.
Docente - Investigadora Educativa.
Doctora en Cs. de la Educación, Magíster en Desarrollo Curricular y Licenciada en Relaciones Industriales.