• Wed
  • Dec 24, 2014
  • Updated: 5:15am
Column
PUBLISHED : Monday, 13 May, 2013, 12:00am
UPDATED : Monday, 13 May, 2013, 9:54am

Learning Curve: the tyranny of predicted grades

As students sit their exams for the International Baccalaureate (IB) and International General Certificate for Secondary Education (IGCSE), I am reminded of the hundreds of predicted grades I have given out in my teaching career.

How accurate have I been and how have my predictions affected my students’ admissions to universities?

Researchers Nick Everett and Joanna Papageorgiou, from Britain, investigated the accuracy of predicted GCE A-level grades as part of the Universities and Colleges Admissions Service admissions process. They report that 51.7 per cent of predictions were accurate, 41.7 per cent by at least one grade. Only 6.6 per cent of all predicted grades were under-predicted.

“A” grades were most accurately forecast (63.8 per cent accuracy) compared to 39.4 per cent of “C” grades.

What other factors influenced the accuracy of my predictions? Everett and Papageorgiou report that gender plays a part in predicting grades. Female applicants were more likely to achieve their predicted grades and were more likely to be accurately predicted than male applicants.

And teaching to a higher socio-economic group at an independent school has apparently made my predictions more accurate.

As an educator, it is much easier to predict a seven for IB and an A* for IGCSE based on school assessments when students marks exceed the mark boundary for those grades.

But a five can easily change into a four or a six after moderation or a good/bad final exam performance. Inaccurate predictions do not affect admission to universities in the US, because SAT scores and extracurricular activities are factored into the process.

But grade predictions have a significant impact on students applying to universities here and in Britain. Consider a student from my higher level biology class who secured admission to medical schools on an optimistic prediction of 38 points for his IB.

If he fails to achieve this grade, he will need to accept admission to an institution that was his backup choice, or seek admission through clearing, which is getting increasingly competitive.

If he had been accurately predicted to attain points lower than 38, he would not have been eligible to study medicine. That would have allowed him to apply to a range of universities that would offer subjects leading to an alternate career path, to which he would have probably given more thought.

But for the institution he applied to, several optimistic predictions result in missing out on students of a higher calibre, who would actually have met their predicted grades. This leads to a low attainment rate for the university.

Consider another student forecast to achieve 34 points, which means he cannot consider studying medicine. He will be forced to apply for programmes leading to his alternate career choice. But if he eventually does secure 38 points, the pessimistic prediction will probably cause him to miss out on his career of choice.

The only indicators of academic ability for students taking the Hong Kong Diploma of Secondary Education, or those studying the Middle Year Programme before taking the IB diploma, are their predicted grades. Students taking IGCSE before the IB or A-levels have an advantage as they can present their IGCSE grades to universities in support of their predicted grades.

Late bloomers, and students who started studying seriously after their predicted grades were declared, are disadvantaged.

We need a fairer university admissions system based on actual exam results, not unreliable predictions.

Anjali Hazari teaches IB and IGCSE biology at the French International School

Share

For unlimited access to:

SCMP.com SCMP Tablet Edition SCMP Mobile Edition 10-year news archive
 
 

 

This article is now closed to comments

donniemcm
"They report that 51.7 per cent of predictions were accurate" conclusion is : they have 1 chance out of 2 to predict well if you pass or fail i.e. this is BS
 
 
 
 
 

Login

SCMP.com Account

or