Has a GCSE grade C/4 lost its value?
By IOE Editor, on 16 March 2021
By John Jerrim
With exams being cancelled again in 2021, concerns have resurfaced over there being rampant “grade inflation”. This was thought by many to be a perennial problem throughout the nineties and noughties as GCSE and A-Level pass-marks raised year-upon-year. Yet, when the Conservatives came to power in 2010, this was something that they vowed to end.
Indeed, some argued it was this obsession over avoiding grade inflation and maintaining standards that ended up leading to the disaster we saw with the awarding of examination grades last summer.
But how have GCSE standards really changed over time? This blog takes a look.
How can we do this?
To examine grade inflation over time I look at average PISA mathematics scores by GCSE mathematics grades, covering examinations taken between 2007 and 2019. PISA is taken just six months before GCSEs, and provides an independent benchmark that is meant to be comparable over time (and hence not affected by problems of grade inflation). (Though readers should keep in mind some of the caveats I have written about PISA previously.
This takes us through to 2019. But we know a lot of grade inflation also occurred last year due to the disaster that came with cancelling examinations. I have produced a rough estimate of the impact that this has had in terms of PISA scores and so also include this in the results (gory details of how I came up with this estimate can be found at the end of the blog).
Together, this allows us to look at changes since 2007 in the “value” (in terms of PISA points) of England’s GCSE “standard pass” (grade C in old money, grade 4 in new money).
A sustained fall in value
Figure 1 provides the headline results, comparing average PISA mathematics scores for those who achieve England’s 4/C grade standard pass in mathematics (green line) to the average for England as a whole (pink line) and across all OECD countries (blue line).
Figure 1. The decline in PISA scores associated with achieving grade C/ 4 in GCSE mathematics in England since 2007
Notes: Grade C figures in 2012 based upon state school pupils only (illustrated by dashed line). Data for 2020 estimated from change in the grade distribution between 2019 and 2020. Year running along horizontal axis refers to the year GCSEs were taken by the PISA cohort.
There has been a clear and sustained fall in the value of England’s GCSE standard pass (grade C/4) over time. Back in 2007, hitting the floor target (grade C) was equal to the OECD average (495 points). Yet this value has got gradually eroded away, down to around 470 after Michael Gove’s time in charge and down to around 455 points for a grade 4 in 2020.
To put this into context, the OCED would equate this 40-point fall in the value of England’s floor target between 2007 and 2020 to a drop of more than one whole year of schooling.
Table 1 below provides further contextualisation of these results. This takes the PISA 2018 mathematics rankings and inserts the value of England’s floor target since 2007 (for each year with data or an estimate available). I have also thrown in the “value” of grade 5 in 2019 and 2020 (estimated) for good measure.
Table 1. The position of England’s C/4 and 5 grade within the PISA 2018 mathematics rankings
|Country||PISA score||Country||PISA score|
|Hong Kong (China)||551||Australia||491|
|Chinese Taipei||531||OECD average||489|
|Netherlands||519||GCSE C (2010)||484|
|Belgium||508||GCSE C (2016)||473|
|Grade 5 (2019)||506||Malta||472|
|England average||504||GCSE grade 4 (2019)||467|
|United Kingdom||502||GCSE C -state pupils (2013)||463|
|Germany||500||GCSE grade 4 (estimated) 2020||456|
|France||495||.+ 33 further countries….|
|Grade 5 (estimated) 2020||495|
|GCSE C (2007)||495|
Children who got a grade 4 in mathematics in 2020 had roughly the same skills as the average young person in countries like Turkey and Ukraine. This is around 16 places lower in the PISA rankings than a child who got a C grade back in 2007 (roughly equivalent to countries like Australia, New Zealand and France).
Table 1 also illustrates how a grade 5 in 2020 now has almost exactly the same value as a C grade did back in 2007. Post-pandemic, this might justify the 5 grade becoming the new floor target young people are expected to achieve.
A tale of two policy mistakes
The devaluation of England’s floor target has really stemmed from two key mistakes.
The first was the miscommunication of what the floor target and expected standard was when the new 9 to 1 grading system was introduced – and whether grade 4 or 5 should be considered the “pass” mark.
The second was in the response to the cancellation last summer, which led to the eventual reliance upon (inflated) centre-assessed grades.
Now, only time will tell whether we will see yet more grade inflation this summer, as some suggest. But, if we do, then a conversation may be needed as to whether existing floor targets continue to hold sufficient value.
Table 2 below provides three pieces of information:
- Column (i) = Average PISA scores by GCSE grades in 2019.
- Column (ii) = GCSE mathematics grade distribution in 2019
- Column (iii) = GCSE mathematics grade distribution in 2020.
To estimate grade inflation in terms of PISA scores, I first multiply average PISA scores at each grade by the percent of pupils who achieve each grade. This is done for both 2019 (second column from the right) and 2020 (the righthand-most column). The values in these columns are then summed together, giving a kind of “weighted average” for 2019 (486.5) and 2020 (497.6). The difference between these two values – 11 points – is my estimate of the impact of the 2020 grade inflation. Note that I treat this as a single value that causes a monotonic shift in the distribution, affecting all parts of it equally (i.e. I do not allow for any potential differential inflation at different grades).
Table 2. Estimating the value of grade inflation between 2019 and 2020 in terms of PISA points (mathematics)
|(i) Average PISA score||(ii) GCSEs 2019||(iii) GCSEs 2020||column (i)* column (ii)||column (i)* column (iii)|