Answer? Law and astronomy, although there are very few entries each year. The much bigger issue is that GCSEs in modern foreign languages are graded more severely than other subjects.
Just before Christmas, Ofqual published a set of very interesting working papers about inter-subject comparability and subject difficulty in GCSEs and A levels. The conclusion is unequivocal. Grades in some subjects are higher (or lower) than in others.
In this blog, we look at the difficulty of different subjects at GCSE as we’ve written about A levels previously. As some astute readers noted in response to that blog, strictly speaking what we are actually discussing is severity of grading rather than subject difficulty. Some subjects are graded more (or less) severely than others.
Let’s begin by grabbing some data. We use results in full course GCSEs entered by all pupils who reached the end of Key Stage 4 nationally in 2015 in all institutions including independent schools, special schools and alternative provision. A pupil’s first entry is used in cases where a subject has been entered more than once.
Assigning grades on the basis of latent ability
Underlying our analysis is a statistical model that assumes that grades in GCSEs are awarded on the basis of both the ability of pupils and the difficulty of each subject. As we wish to compare different subjects, we assume that the measure of ability does not vary between subjects but the level of difficulty does.
A measure of each pupil’s latent ability is estimated from the data. This scale can then be sliced up into grades. Chart 1 shows the cut-points of our notional ability scale for a subject of average difficulty. This suggests, as we have discussed previously, that intervals between grades are not equal. In other words, the difference between G and F is not the same as the difference between B and A.
Chart 1: GCSE grade cut-points on the latent ability scale
Variation in difficulty between subjects
Table 1 below presents all subjects with at least two thousand entrants amongst the 2015 national KS4 cohort in ascending order of difficulty (or grade severity, if you prefer).
Pupils entered for Polish, Chinese, Arabic, Italian, Urdu and art & design tend to achieve better grades in those subjects than might be expected from their latent ability. This makes sense: these subjects tend to be entered by pupils who are either native speakers or who have an aptitude for art & design. In other words, they tend to be more self-selecting than in other subjects and so it is little surprise that results tend to be higher. I very much doubt that Polish would remain the least severely graded subject if large numbers of pupils who were not Polish started to enter it.
By contrast, entries in law, astronomy, applied engineering, general studies and German tend to be more severely graded. Perhaps more worrying is the difficulty associated with the three most common modern foreign languages taught in schools: French, German and Spanish.
The scale on which subjects are ranked in Table 1 is not particularly meaningful so Chart 2 shows what it means in practice for a selection of subjects. Here we show the difference in percentages of A*-C and A*-A grades awarded compared to a subject of average difficulty.
Chart 2: Differences in percentages of A*-C and A*-A grades awarded compared to a subject of average difficulty
So what?
For a start, Ofqual deserves credit for launching this debate. But then we have to ask whether we should have consistency of grading between subjects. Is it even technically feasible? If we cannot have a perfect balance between subjects is there anything we could do to get a bit closer?
Even if we “fix” the current inconsistency between subjects, there is no guarantee that grades won’t diverge again in the future. Perhaps French used to be no more severely graded than maths. The difference we see today may be the product of changes in teaching and attitudes.
Or should we just live with inconsistency? If so, we have to understand the consequences. Schools’ curriculum choices influence their performance measures. At present the more that schools offer MFL the more they are effectively being penalised.
Table 1: Estimated GCSE subject difficulty, pupils reaching the end of KS4 in 2015
Subject | Change in log-odds of higher grade compared to maths | Entries |
Polish | 5.24 | 3768 |
Chinese | 2.49 | 3153 |
Art & Design (Photography) | 1.51 | 22006 |
Arabic | 1.50 | 3035 |
Italian | 1.45 | 3854 |
Urdu | 1.21 | 4087 |
Art & Design (Textiles) | 0.84 | 6749 |
Art & Design (Fine Art) | 0.64 | 50999 |
Art & Design | 0.62 | 86018 |
Expressive Arts & Performance Studies | 0.48 | 3151 |
D&T Textiles Technology | 0.47 | 24108 |
Art & Design (Graphics) | 0.45 | 6874 |
D&T Food Technology | 0.30 | 38263 |
Film Studies | 0.30 | 6209 |
Media/Film/TV Studies | 0.26 | 51402 |
Home Economics: Child Development | 0.26 | 18136 |
Performing Arts | 0.21 | 5937 |
Religious Studies | 0.11 | 267864 |
English Literature | 0.06 | 405670 |
Drama & Theatre Studies | 0.06 | 70655 |
English Language & Literature | 0.06 | 64535 |
Mathematics | 0.00 | 542020 |
Physical Education/Sports Studies | 0.01 | 110391 |
Social Science: Citizenship | -0.02 | 20731 |
Information & Communications Technology | -0.02 | 99150 |
English Language | -0.07 | 298688 |
Physics | -0.12 | 122528 |
Health & Social Care | -0.12 | 7079 |
Chemistry | -0.13 | 123333 |
Latin | -0.13 | 8337 |
D&T Resistant Materials | -0.14 | 50826 |
Dance | -0.15 | 11953 |
Biology | -0.17 | 126042 |
Home Economics: Food | -0.22 | 8628 |
Methods in Mathematics | -0.34 | 12198 |
Office Technology | -0.41 | 13340 |
Applications of Mathematics | -0.43 | 12097 |
Sociology | -0.43 | 21274 |
D&T Electronic Products | -0.44 | 7862 |
Science (Core) | -0.45 | 366238 |
D&T Product Design | -0.48 | 37750 |
Music | -0.49 | 43502 |
Additional Science | -0.53 | 326679 |
Statistics | -0.57 | 51647 |
Geography | -0.57 | 210919 |
History | -0.63 | 228569 |
Business Studies | -0.72 | 73819 |
D&T Graphic Products | -0.73 | 31708 |
D&T Systems & Control | -0.85 | 2969 |
Spanish | -0.99 | 84846 |
Classical Civilisation | -1.10 | 3971 |
Psychology | -1.21 | 15897 |
Computer Studies/Computing | -1.28 | 33378 |
Economics | -1.29 | 9414 |
French | -1.34 | 150702 |
Humanities: Single | -1.35 | 8326 |
German | -1.44 | 52621 |
General Studies | -1.71 | 9302 |
Applied Engineering | -1.88 | 6257 |
Astronomy | -2.46 | 2300 |
Law | -2.47 | 2196 |
Nice piece. I’d like to know what the most useful GCSEs are for life after school. I.e the subjects that most people use in their jobs and daily lives.
In my personal unbiased opinion:
Maths (not so much use of the content but how well regarded a high grade is)
English Lit/Lang (mainly writing a structured essay, which is a useful skill)
Business Studies (a lot of very useful content in the course itself, but it isn’t exhaustive. It’s also poorly regarded)
Physics (if you get an engineering or tech job, this could be very helpful)
What I find frustrating, is that is not a new issue.
Previously analysis of exam performance looked at ‘weighted residuals’ as well as ‘raw residuals’
We were able to compare how pupils’ performance in a subject compared to all pupils nationally who took that subject
So I could see (for example) that although my colleague teaching Basket weaving had a residual of -0.2 (each pupil they taught did 0.2 grades worse with them than their average grade in all subjects) it turns out that nationally basket weaving had a residual of -3.2. So in fact their kids had done 3 grades better than we might expect.
But for some reason those national residuals are just not available (if you know different then please do let me know where to find them) so we are reduced to a simplistic analysis of comparing performance between subjects in the same school. Which will always sum to zero! So SLT under pressure for league tables will prioritise subjects which gets grades which look good, and not those which might actually give kids life chances.
Now we have 2018 results, would Education Data Lap please update this blog? Are the reformed GCSEs in languages still harshly graded? I suspect they are having seen the correlation chart at the FFT roadshow last week. Why is this so??
Hi Jo, thanks for the suggestion. We do plan to revisit this topic – but probably after the 2019 exam series, by which point almost all GCSE subjects will have been reformed.
I await this with baited breath as a teacher of MFL who can not persuade my Leadership that it is harder to get a “high” MFL grade than the other Ebacc subjects. According to chart 2, French is -0.62 compared to History. That (almost) 2/3 of a grade makes a hell of a difference in results with my pupils.
Hi, I’m 16 and currently sitting the 2019 GCSE exams and came here to see what the hardest subjects are, personally I find maths the hardest but I am good in English, business and biology. I can inform you that lots of the exam board use more complicated real-life associated context nowadays which can make things harder. Please update this post for the recent exams!