In an inconspicuous semi-detached house, on a leafy residential street in Cambridge, decisions are being made that will shape the futures of tens of thousands of teenagers. Over paper cups of coffee and as-yet-unopened packets of biscuits, four senior examiners are deciding, in effect, how many students in England will pass their history GCSE - and how many will get an A.
This is no clandestine, illegal operation: the house is used as meeting room space by OCR, one of England's main exam boards, and the business is official. The examiners are gathered in one of the small upstairs rooms, reading a sample of students' already-marked "controlled assessment" papers to help them decide where to set the GCSE grade boundaries.
Later, their recommendations will be reviewed by an even more senior group of examiners, who will look at data about the cohort of students - and analyse the recommendations of other groups of OCR examiners who have looked at different papers - before finalising the boundaries.
To a casual observer, this process of setting GCSE grade boundaries would appear to be the same as it has for decades, but in recent years it has in fact undergone a major change. Historic data now plays a more important role than ever before. Current grade boundaries are to a large extent determined by the performance of a cohort of pupils in Year 6 Sats tests, and by previous cohorts' GCSE results, through an approach known as "comparable outcomes". The examiners' own judgements now carry less weight.
Of course, this tweak to the system, to tackle what was seen as grade inflation, is not the only recent change to GCSEs. Rather, it is part of a broader effort by the government and other bodies to make GCSEs tougher. From next month, schools will start teaching completely new GCSEs that have been explicitly designed to be more demanding.
Pupils will also have to adjust to the replacement of A*-G grades with a numerical system. Far fewer candidates will be awarded the top grade 9 than currently achieve an A*. And the government has set the bar for a "good pass" at grade 5, which is equivalent to a low B rather than a C.
The theory is that tougher GCSEs mean more capable, knowledgeable students. A rise in standards should, it is argued, follow. But can it really be that easy?
The harder they come
Conservative politicians seized on the theme of tougher exams soon after coming to power in 2010, with Michael Gove, who was education secretary, announcing in the autumn of that year that he wanted a "more challenging curriculum" and a "qualifications system that restores standards".
Much of the ire was focused on grade inflation, with many - employers chief among them - suggesting that an A grade was not what it once was. Twenty-four years of continuous GCSE grade increases had been attributed by ministers in the New Labour years to better teaching and greater investment in schools. But employers - led in their complaints by the Institute of Directors, which throughout much of the 1990s and 2000s condemned "endemic and rampant grade inflation" - were sceptical.
There were admissions within the exams world that things may have gone awry, too. Tim Oates, head of research for Cambridge Assessment, which owns OCR, went public in 2010 with his concerns that a culture of "giving the benefit of the doubt to pupils" had resulted in "subtle grade inflation".
Ofqual was tasked with tackling this "grade inflation" and, in October 2011, Gove told a summit hosted by the exams regulator that he was relaxed about the prospect that grades might fall as a result of the changes it needed to make.
And fall they did: in August 2012, those 24 years of continuous GCSE grade increases came to an end. The proportion of students gaining five A*-Cs including English and maths fell by 0.4 percentage points to 58.6 per cent.
The solution that Ofqual had come up with was an extension of what exam boards had been doing for some time: using data on pupils' prior attainment to inform grade boundaries. But under a more formal and prescriptive version of this "comparable outcomes" approach, such data went from being useful background knowledge that informed examiners' professional judgements to being the main determinant of where grade boundaries should be set.
The net result is that a similar proportion of students will receive a given grade each year, and it has become harder for grades to rise overall.
`Raising the bar'
The approach fulfilled demands for an end to grade inflation, but it created a new problem: even if schools across the country dramatically improved their teaching and students worked harder than ever before, their efforts - at a national level - would not necessarily be rewarded with better results.
In theory, higher standards could still lead to rising grades. If the expert examiners at the house in Cambridge can make a rock-solid case that higher grades than those dictated by the "prediction matrix" are deserved, then the system is supposed to allow it. But the question is: how can that case be proved without an independent measure of standards, separate to GCSE grades?
Ofqual is hoping to solve that problem with its new "national reference test" (see panel, opposite). Many are pinning their hopes on this test as the solution to the improvement dilemma.
Ministers seem confident that schools will cope with the changes to grade boundaries and the wider reforms to make GCSEs tougher. "I think what we have seen when we have made changes, when we have raised the bar, is that actually schools and colleges have risen to the challenge," education secretary Nicky Morgan told a conference at Wellington College in June. She was responding to a question about whether the introduction of more difficult GCSEs would cause schools' performance to decline.
Out of schools' control
But schools are not convinced by her argument. Susan Cousin, principal of Yewlands Academy in Sheffield, says the strict use of comparable outcomes to set grade boundaries means there is no longer a guarantee that a given level of performance will lead to a particular grade - and that means trouble for teachers and students.
"Teachers want to know what needs to be delivered to attain a certain grade," she says. "What is unforgiveable to me is that the goalposts are moved after students have taken their exams. To me, if 90 per cent of students reach a standard that has been defined as an A grade, those students should all get an A."
David Blow, headteacher of the Ashcombe School in Surrey, says much of his work has been to make sure his students - and teachers - do not panic about the scale of the changes. "I have been trying to take the stress out [for students] by saying, let us concentrate on what is within our control, such as engaging with lessons and working hard," he explains. "Whether their work is fairly reflected [in grades] will depend on how the system operates, and that is outside of our control."
At the Haywood Academy in Stoke-on-Trent, pupils on GCSE courses have been sitting mocks in the exam hall every six weeks to prepare them for the new high-stakes tests. "If you give a child from a tough background just one mock exam, it is not enough to build their resilience up to cope with the new GCSEs," says headteacher Carl Ward.
But Ward is still worried about the impact of the reformed GCSEs, particularly because the pass mark is higher than before. "It's easier for a child from a middle-class background to get a B than it is for a child from a deprived background," he says. "Ramping up the boundaries makes it harder for children from tough backgrounds to get the accepted pass grade."
The elusive pass mark
That accepted pass grade, however, may not be all it seems, causing even more issues for schools. Ministers' insistence on using grade 5 as a "good pass" might not be as tough as it sounds: after all, the pass mark is not set out by law.
"The government cannot tell employers and universities what grades they should require," says Michael O'Sullivan, chief executive of Cambridge International Examinations, sister board of OCR. "I think it [grade 5] is unlikely to be adopted wholesale as a pass mark."
Since the lower grade 4 will be statistically linked to the C grade, employers and universities might choose to require this. Doing so would avoid the prospect of legal challenges if, say, a firm asked for grade 5 from job applicants taught in England and grade C from those taught in Wales, where the 1-9 system will not be introduced.
Another problem is that making GCSEs harder is not proving an exact science. The exam boards are seemingly finding it difficult to pitch questions at the "right" level, with serious consequences for teachers.
The clearest sign of this came with the debacle this spring about exam boards' sample papers for the reformed maths GCSE. Three of the boards - OCR, Pearson's Edexcel and WJEC Eduqas - were told by Ofqual to make their higher-tier papers easier. The fourth, AQA, was told to make its foundation paper harder.
It was an embarrassing situation for Ofqual because the regulator had already approved the papers for teaching when some of the boards complained that AQA's materials were too easy. This prompted an in-depth investigation that led to all boards being told to make changes.
O'Sullivan says the fiasco illustrates "what tends to happen when you rush reforms through", adding that one of the reasons for the educational success of places such as Hong Kong and Singapore is that they have introduced reforms after careful piloting.
But the problems also raise the question of whether England's model, in which exam boards compete in a regulated market, will limit any increase in the difficulty of the new GCSEs. Without certainty that all GCSEs are at exactly the same level of difficulty, is there a risk that standards will be levelled down to match the "easiest" board?
Rod Bristow, president of Pearson UK, does not think so. "Exam boards do not and should not compete on standards," he says, adding that it is "important for the credibility of awarding organisations" that they are seen to be setting their papers at the required standard. "Ofqual is focusing very hard on [standards], and I do have confidence that we will get to the right place," he concludes.
Others are less confident of a positive outcome to the GCSE changes. A key complication in making GCSEs tougher is the fact that sample papers are trialled on current students, who have not been prepared for new qualifications. If those students score very low marks on the sample papers, Ofqual could treat this - as it has with the maths GCSEs - as a signal that the papers must be made easier.
Professor Alan Smithers, an assessment expert at the University of Buckingham, says this is the wrong approach. "If you test [the new papers] out on pupils who have been taught the old GCSEs, then naturally they are going to find them very difficult," he told TES in May, in the wake of the maths fiasco. "We may be lowering what we ask for in the new GCSEs to cater for people who have been taught under the old regime."
Reforms v resources
Clearly, plans to create tougher exams are not running as smoothly as they could be. But if the end result is a raising of standards, surely the teething problems are worth the pain? And yet, the whole concept of making GCSEs tougher to boost standards has its detractors. Blow, for example, says that the government's argument for change employs "false logic".
"It implies that schools are not pushing pupils at the moment," he says. "But schools are already working incredibly hard, and the quality of teaching and learning is high."
Once the new system comes in, Blow adds, teachers and students will work "flat out" to cope with the changes. However, he says, there is no evidence to suggest that tougher tests by themselves will raise standards, and even if a school does improve standards, the use of comparable outcomes means it can only be rewarded with better grades at the expense of other schools' grades declining.
Members of the exams world are also doubtful. O'Sullivan says he supports the switch from modular to linear exams, and the introduction of more content to the maths GCSE, but he is concerned that too much emphasis is being placed on exam reform as a way of improving standards.
"The most promising thing to focus on to raise attainment in schools would be the teaching resource," he says. "It is quite striking when you compare the UK with other countries that the retention of teachers is quite poor.It seems this is not being addressed with the determination it could be, and excessive reliance or hope is placed on changing the assessment system and increasing the demand of assessments as the main way of pulling up standards."
Bristow shares his concerns, although he too supports many elements of the reform programme and believes qualifications should be rigorous and internationally benchmarked, as is the intention of the new English system.
"If you are only worrying about test outcomes, it can become a rather narrow measure of success and can, in the end, lead to the examination tail wagging the teaching and learning dog, and that must not happen," he says. "If the teaching and learning is purely defined in terms of the exam, the risk becomes that children do not learn everything they should be learning, but rather learn how to pass an exam."
Clearly, the link between tougher GCSEs and raising standards is not as clear cut as some may like to think. But if the measures do prove effective, that will pose an important question for the future: if standards need to be continually raised and exams prove an effective means of facilitating that rise, how hard might GCSEs eventually become?
National reference test: a silver bullet?
If GCSE grades among a cohort of pupils are tied each year to the previous cohort's results, how can genuine improvement be recognised? That is the question that the "national reference test" has been designed to address.
The test, being developed by the National Foundation for Educational Research, is to be sat by a sample of about 300 schools every March, starting in 2017. If students perform better in one year than in the previous year, exam boards could treat this as a sign that some "grade inflation" would be fair, allowing them to set the grade boundaries lower than they otherwise might.
To make sure pupils' performance on the national reference test can be compared from year to year, exams regulator Ofqual is planning to keep the questions in the national reference test "largely the same" each year.
But that raises its own potential problem: how to stop those questions becoming widely known and, of course, shared on social media?
Find news, advice and stories from students and teachers at www.tesconnect. comgcseresults