GCSE and A-Level results will be decided by statistical modelling – NOT teacher’s predicted grades – in major government U-turn on exams
- Exam regulator Ofqual said statistical modelling will be used to calculate grades
- Model will factor previous grades of pupils and former students at same school
- Ofqual said teachers’ predicted grades submitted in March will also be used
Statistical modelling will be used to determine the majority of this year’s A-level and GCSE results as opposed to predicted grades from teachers.
Exam regulator Ofqual announced the government u-turn after concerns regarding the reliability of teacher-predicted grades were raised.
The new statistical model will take into account a number of factors, including pupils’ previous attainment, results of previous students at the same school and the predicted grades teachers submitted in March.
A new statistical model will be used to generate most of this year’s A-level and GCSE results as opposed to teacher’s predicted grades (Stock image)
The latest assessment update comes as earlier this week Ofqual said a ‘substantial number of students’ will be given lower GCSE and A-level grades than their teachers wanted.
Data from the government office suggested if teacher-assessment grades were not standardised, this year’s results would be around 12 percentage points better than in 2019 at A-level and nine percentage points at GCSE.
Ofqual said it was ‘not surprising’ that the grades calculated by many schools and colleges were more optimistic, as teachers ‘naturally want to do their best for their students’.
Despite all exams being cancelled due to the pandemic this year, hundreds of thousands of children will receive results next month which Ofqual have assured will be fair and will not penalise anyone.
Exam regulator Ofqual insists the results will be fair and will not penalise any students (Stock image)
When requesting predicted grades earlier this year, the Department for Education instructed teachers to factor in previous mock results and how the children had performed that year.
These results were then to be going to be combined with other data to formulate the grades but sources told The Sunday Telegraph: ‘If Ofqual are able to use statistical standardisation to produce grades, then in effect the teachers’ predicted grades serve little or no purpose’
To try and find the best model for generating this year’s grades, 12 different statistical models were tested.
The exam regulator said the predicted grades submitted in March were an ‘important competent’ in the models and helped check for ‘quality assurance’ while testing out the different methods.
Predicted grades will play a bigger role for pupils at new schools which do not have previous data to drawn upon and at smaller schools.
Despite assurances from Ofqual that results will be calculated fairly, some headteachers believe pupils who do not receive the grades they had hoped for will feel like victims of the process.