Are improving examination results the product of better performance or less demanding assessment? John explores the theory and practice of grade inflation.
When I was a young Oxford don interviewing admissions candidates, three A grades at A-level was exceptional, but it is now the norm. Yet few of my colleagues believe that the students they teach are brighter, or better prepared. So when, yet again, more school students than ever before secured A grades this year, I was sceptical. But facts, not personal anecdotes, are what matter.
In 2000, the UK’s Qualifications and Curriculum Authority commissioned a report from a panel chaired by a distinguished American educationist. The panel’s conclusions are often quoted in support of the claim that standards have been maintained. But this interpretation is unjustified. The panel recognised that examination procedures were careful and conscientious. But they noted that they had no evidence to judge what had happened to grading standards and urged that more effort should be made to collect such evidence.
As a result, the QCA has asked assessors to review recent and older scripts in various subjects, including mathematics. Mathematics is probably the best subject to use for comparisons. Marking is relatively objective and the skills appropriate for school students change little over time. The judges concluded that things were just fine. The London Mathematical Society, by contrast, claims to observe a marked fall in mathematical skills. But their report is old – published in 1995. These conflicting statements of personal opinion tell us little.
Grades might improve even if students did not get any better, nor exams any easier. The same student is more likely to get a good grade in media studies than physics. If media studies rises in popularity relative to physics, as it has, the proportion of good grades will increase. If schools teach to tests – performance tables encourage them to do so – then students might become better at taking exams even if their skills and knowledge have not improved.
So what has actually happened to student skills? The nearest to objective measures are international tests of student attainment, such as the Programme for International Student Assessment study sponsored by the Organisation for Economic Co-operation and Development; and the Trends in International Mathematics and Science Study comparisons made by the International Association for the Evaluation of Educational Achievement. The earliest such assessment goes back to 1995. Among developed countries, England is in the middle of the pack.
Unfortunately, the response rate in England to these surveys is poor. Perhaps teachers feel, with some justification, that there are already enough tests. The authors are therefore unwilling to comment specifically on English trends. But achievement standards across the world have remained broadly constant across the survey period and nothing in these studies suggests that England is any different. The most likely cause of improving examination performance is grade inflation – it becomes slightly easier each year to get an A grade.
If you have ever filled in a standardised evaluation form, you know that about half the people assessed, necessarily, are below average. But to say someone has below average intelligence is to call them an idiot; to say their integrity is below average to imply they should be in prison. If you were to complete such a form honestly the subject would likely be unemployed.
The best way for an instructor to be popular with students, an examination board to attract candidates, or for an examiner to feel good about his or her job, is to temper rigour with accommodation. Most people want to please. I have been the grumpy examiner who resists attempts to push grades across the borderline. I feel like a heel and so I should. The desire to give people chances they may not altogether deserve is an admirable human trait. It is just not the trait you want in examiners.
The only way to offset a tendency to grade inflation is through the cold logic of statistical distributions. People are nicer than computers. That is why education needs warm-hearted people to inspire us but cold-blooded computers to record the outcomes.
