A-level grade predictions mostly wrong
Most A-level grade predictions made by teachers are incorrect when the final results are published, according to data revealed by an exam body.
Predicted grades are an important part of the university application process.
Only 48% of predicted grades were correct for pupils taking the OCR exam board's A-levels in summer 2012, in England, Wales and Northern Ireland.
Almost one in 10 of the wrong forecasts, more than 15,000 entries, were incorrect by more than one grade.
For this one exam board, there were more than 186,000 incorrect forecasts for that summer's A-levels.
These teacher forecasts were gathered by the OCR board in May, a few weeks ahead of the exam season.
As such they should have been more accurate than the January predictions used in the university admissions system.
The findings could re-ignite questions about the fairness of a prediction-based university application system, before pupils know their results.
Although, the exam body says, it means more than 90% of forecasts were correct within a margin of one grade.
The A-level statistics published by Cambridge Assessment, OCR's parent body, show that teachers were much more likely to be over-optimistic about results - with 39% over-predicting, compared with 13% whose predictions turned out to be lower than the actual grades.
Independent school teachers were the most accurate in predicting the right grade, followed by grammar schools, academies, sixth form colleges and comprehensives.
The lowest level of accuracy was for teachers in further education.
Predictions from further education colleges, as the least accurate at both ends of the scale, were both the most likely to be over-optimistic and the most likely to be wrongly pessimistic.
Independent and grammar schools were the toughest in terms of predictions, with the lowest levels of over-optimism.
Teachers were much stronger at predicting the highest grades, A* and A, with more than 60% accuracy. For C-grades, only 47% had been correctly predicted.
But the figures also show that over a range of subjects, when a pupil is forecast three A-grades or better, only 35% of these predictions prove to be correct.
Oxford University said that there were other factors considered in applications, as well as predicted grades.
"We look carefully at a range of factors when considering a student's application, including GCSE scores, aptitude test results, interview performance, teacher reference and Ucas statement, so that we are not reliant on predicted grades to select the very best students," said admissions spokesman, Matt Pickles.