Skip to main content

Blog The Ofqual blog


New A level maths in 2019

Posted by: , Posted on: - Categories: A levels and GCSEs, Exams

For some years, maths has been the largest entry A level in England. In 2018 around 90,000 students took A level maths, with just over 15,000 also taking A level further maths. The introduction of the new A level was deliberately included in those subjects for first teaching in 2017, so that the students taking the new A level would first have studied the new 9 to 1 GCSE maths.

Some schools and colleges entered students for the new A level last summer and we explained last year why maths was available after only one year’s teaching. As a result, there are past papers and grade boundaries available from 2018, but we urge caution in relying on those 2018 grade boundaries. Only a few thousand students, mainly those in year 12, took the new A level in 2018 and they tended to be very able mathematicians who were also studying A level further maths. As a cohort, those few thousand students will have performed very differently to the tens of thousands that we expect to see this summer.

In recent years, as we have seen the introduction of new A levels and GCSEs, we have cautioned about the likelihood of increased variability in results for individual schools and colleges. But that’s generally not what we’ve seen – schools and colleges appear to have adapted very well to the changes, and results, both overall and at school/college level – have been stable.

However, the changes to maths are significant, and they follow a long period of stability in the specifications, much longer than for other subjects. The main changes are as follows:

  • As for all other reformed subjects, the new A level is decoupled from AS, so AS units do not count towards the A level grade. Since the legacy maths was a six-unit qualification, this means students taking the new A level will sit fewer exams.
  • However, the move to a linear structure is more of a change in maths than in other A level subjects, as A level maths students taking the legacy specifications could choose to take 4 AS units and 2 A2 units to achieve their A level (in other 6-unit A levels, students would have taken 3 AS units and 3 A2 units).
  • The new A level maths has entirely compulsory content, covering both statistics and mechanics, so some teachers might be teaching material they are less familiar with.
  • The complex structure of the legacy specifications, with many optional AS and A2 units, some of which could be used for both maths and further maths, has been simplified. A level maths is now separate to further maths and the units for the respective qualifications are different.
  • Previously, students often sat some units more than once, and unit results were combined to maximise the grade (or grades where students were also taking further maths) achieved. In all the new linear A levels, students’ grades are determined by the number of marks obtained in their exams taken at the end of the course.Image of fountain pen, calculator and ruler set on squared paper

All of the above changes make it challenging for senior examiners to judge the quality of student work compared to previous years. And we know from our research into the sawtooth effect in new qualifications that student performance tends to dip slightly in the first years of a new qualification. That’s likely to be only a small number of marks, but it makes the senior examiners’ task of setting grade boundaries even more challenging. So our approach this summer in the new A level maths will be the same as for the first awards of the other reformed A levels. Exam boards will use predictions based on the cohort’s prior attainment at GCSE so that if this year’s cohort is similar, then we would expect the profile of grades to be similar to previous years.  We do this in order to be fair to this year’s students and to make sure that exam boards’ standards are aligned.

If you would like to talk to Ofqual about any of the issues raised in this blog, please contact us at

Sharing and comments

Share this page


  1. Comment by Michael Danzelman posted on

    Unfortunately, this blog doesn’t reveal the bigger picture behind the A-level maths reforms. It’s true that the summer 2018 cohort was composed almost entirely of students currently sitting Further maths qualifications and they are, by definition, at the upper end of the ability range. Bearing in mind the grade boundaries for last summer’s Pearson Edexcel papers were 155/300 for a grade B and 184/300 for an A (with a notional 52/100 for a notional A grade in the Applied paper) and corresponding figures of 162/300 and 197/300 for OCR Syllabus A, it’s obvious that these very able students and their teachers struggled to get to grips with the new A-level. If Awarding Bodies have not responded to this in terms of question style and difficulty, it’s likely this year’s grade boundaries will be significantly lower once the full ability range sits these papers (effectively going the same way as GCSE mathematics). Any reference to the sawtooth effect and a “slight dip” in student performance is a serious understatement in the light of current evidence. It’s true that students shouldn’t be disadvantaged by these reforms and so the massaging of grade boundaries is necessary to ensure, as far as possible, parity between the legacy and reformed structures when awarding. However, an unwanted consequence of this is that it causes bunching of grade boundaries (particularly in the Pearson Edexcel Applied Paper last year, for example) and that means small differences in performance can have a big impact on overall attainment. A far bigger potential problem is the negative experience that arises from students sitting exams that they and their teachers find difficult to prepare for and risks damaging the uptake of future cohorts. During the consultation period, MEI warned that tinkering with the existing structure was an unwise move when numbers taking A-level maths were at an all-time high: effectively the message was “if it ain’t broke, don’t fix it”. The Curriculum 2000 reforms serve as a warning of what can happen when this advice is ignored. It’s obvious that these reforms have been driven by political ideology rather than looking at ways the existing system - which was functioning well - could have been improved further. Subject content for the new specification has been hastily put together by the intelligentsia in a “change for changes sake approach”, without any thought for how this might work in the classroom. A lack of a proper pilot study was a big mistake. Thus, we now have a compulsory and fatuous Large Data Set that satisfies senior statisticians (bizarrely no such requirement exists in A level Statistics) and a bloated emphasis on the nebulous concept of modelling to satisfy the engineers. What no teacher or student asked for was an increase in subject content yet, with a reduction in exam assessment from nine hours to six, it is impossible to examine much of this content. These reforms are unlikely to leave students better prepared for a Higher Education course with significant mathematical content: whilst the legacy A-level course provided the structure and clear mileposts that compelled students to master the basics before moving to the next stage, eventually equipping them with the tools of the trade, in many ways the new A-level is barely fit for purpose.

    • Replies to Michael Danzelman>

      Comment by Ian Dexter posted on

      Grade boundaries are set to reflect the demand of the paper, rather than the cohort of students taking the exam. As you recognise, last year’s cohort for the new A level maths papers was comprised of largely year 12 students who were also studying further maths – and as a result, the grade distribution was skewed towards the top grades (more so than for A level maths students as a whole in 2018). Therefore, it does not follow that grade boundaries will be “significantly lower” in 2019 with a full ability cohort. If the 2019 papers are more demanding, then grade boundaries might be a little lower to take account of that, and if the 2019 papers are easier than those in 2018, boundaries might be a little higher. As in 2018, senior examiners in each of the exam boards will review the student work, to satisfy themselves that the standard of work is worthy of an A level grade.

      It is not clear what evidence you are referring to when you suggest that a slight dip in student performance is a “serious understatement”. Our sawtooth research suggests that the number of marks is relatively small, and found that the differences were generally not significant in terms of the standard of student work. You cite GCSE mathematics as an example but the grade boundaries in the new 9 to 1 GCSE maths are not out of line with grade boundaries in the legacy A* to G GCSE maths, taking into account changes in the structure of the two qualifications. In the new 9 to 1 maths, the ‘overlap grades’ between the foundation and higher tiers are higher than they were in the A* to G maths (5 and 4 in the new maths; C and D in the legacy maths). As a result, the lowest grade that is targeted on the higher tier paper is now grade 4, compared to C previously. We have previously explained how the grade boundaries reflect the targeting of the question papers.

      The structural changes made to new AS and A level mathematics qualifications reflect those made to all A levels – to make A levels linear qualifications with AS results not counting towards the A level grade. These changes reflect decisions taken by the Department for Education (DfE) about their policy intentions for all AS and A level qualifications.

      The development of the subject content for the qualification was also a matter for the DfE, who asked universities to set up the A level Content Advisory Board to advise it on the content of this and two other subjects. As for all reformed AS and A levels, Ofqual confirmed that it was able to regulate the final proposed subject content effectively. Ofqual, following consultation, put in place the rules for how the subject content must be assessed. We are keeping the effectiveness of all of the new GCSE, AS and A level qualifications, under review with our focus being on how well the rules we put in place about assessment are working. If we identify any significant issues with how well the assessments in any of the new qualifications are working, we will take appropriate action.

  2. Comment by Michael Danzelman posted on

    Thanks Ian. You say last year’s cohort of Further mathematicians (falling into the top 15 to 20% of the full ability range) who sat the standard A-level paper would be expected to be awarded a higher than usual percentage of top grades when compared to the full cohort. This is a pass rate statistics argument. When you say “therefore it does not follow that grade boundaries will be significantly lower in 2019 with a full ability cohort” it’s not clear where the implication arises. Ofqual’s lengthy and technical investigation into the sawtooth effect rightly points out that at the start of syllabus reform there is a natural dip in student performance which take about three years for students and teachers to recover from as they familiarise themselves with the demands of the new specification. Pearson Edexcel required an aggregate of 352 out of a possible 450 raw marks for an A grade in the summer 2018 legacy syllabus (using M1 an S1 as the option papers here to assure as close a comparison as possible with the new specification) and 184 out of a maximum of 300 on the reformed. Reducing that to percentages and that’s 78% on the legacy and 61% on the reformed where the latter is based on a cohort of the most able. That is a not a “slight dip”. One of your colleagues dismissed any comparison with the legacy and reformed specifications: if so, then that renders the saw-tooth argument at the point of syllabus change as largely irrelevant and we are back to a pass rate statistics argument. The point is that student ability doesn’t really vary much from year to year, the key factor is the difficulty of the paper and that is why UMS was used in the legacy system – to allow for changes in assessment demands.

    Regarding GCSE, the new concept of a “standard pass” (Grade 4 slightly below the old grade C) and a “good pass” (Grade 5 above the old grade C but below a grade B) improves pass rate statistics because the high jump bar has been lowered by the cynical introduction of a Grade 4 as the de facto pass. (I’m sure the well-understood A* to G already gave employers the granular detail they needed to differentiate between candidates anyway). This in turn is the natural response to extra and more demanding content shoehorned into the GCSE in attempt to win public confidence. Net result of the GCSE and A-level reforms is, in terms of pass rate statistics, minimal.

    Something you didn’t respond to was the effect of these reforms on student confidence. Ofqual’s own statistics show that A-level maths entries are down 5% this year and Further Maths by 10%. This against every promise made that these reforms would not make the subject harder and against every piece of advice not to tinker with the current model in a way that might damage the uptake of mathematics post-16. So, whilst it may be too early to tell if this is the start of the trend that was most feared, it has all the hallmarks of the Curriculum 2000 reforms. Blame for this lies squarely with Gove and the DfE but it’s highly unlikely they will admit responsibility for the consequences of these poorly thought out and unnecessary changes.

  3. Comment by isabellammm posted on

    I’m sure we can all agree this new spec Maths A level 2019 has been an utter disaster in terms of both the bizarre difficulty of papers 1 & 2 which bore no resemblance to the last 2 years work & widespread online leaks of ALL 3 papers, a fact which Pearson continue to deny and downplay. Ofqual need to get their act together and say enough is enough - it is very unfair on hard-working students whose university places depend on achieving sound grades. The integrity of this year’s exam has been totally compromised and PearsonEdexcel as an exam board are clearly not fit for purpose.