The furore about this year’s OCR GCSE Computer Science Paper 2 (J277/02) brings into stark relief the gulf between where we are as a subject, and where we need to be, in terms of capability. It’s a time for reflection, not for panic. The responsibility for thoughtful reflection falls on the computing subject leaders.
Subject Leaders (SLs) are the engine-room of the school. Sometimes called curriculum leaders or heads of department, as such these engine-rooms should be given largely unlimited fuel (training, resources and support) and clear guidance (achievable goals and strategic direction) to ensure their success. Often this is lacking, and computing SLs regularly report feeling under-supported: that their subject is poorly understood and under-resourced. Exam results should drive a discussion between the SL and strategic/senior leadership team (SLT) and it’s right that this should ask non-threatening questions of the SL about their department’s capability: do they have the resources, training and support to achieve the best outcomes possible for their students?
Unfortunately, these conversations are not always positive, causing SLs to fear for their careers instead of being able to work together with SLT on the advancement of their subject outcomes. This has never been so clear as in the panic over Thursday’s paper. I hope in this blog to bring some clarity to the situation. I have left out names, and avoided criticising individuals. But to be clear before we start, I think some teachers have gone too far in their response to what was a tough but fair paper, and I give alternative views and a series of actionable recommendations below.
The harsh reality is that the complaints about the paper that inundated Facebook from Thursday evening onwards say more about our ability as a teaching body to prepare our learners for a robust assessment, than they do about the quality or fairness of the exam itself. My unpopular opinion is that the exam was tough but largely fair. Nothing was assessed that wasn’t in the specification, and it seems many teachers have been over-reliant on the practice paper and past papers on OCR’s website and secure “Interchange” area, perhaps leaning on these to guide their curriculum rather too much, instead of trying to cover the whole specification. [Edit: this paragraph previously used the phrase “been caught out teaching to the test” and I have removed and rewritten to be clearer and sound less judgemental].
An over-reliance on questions from the previous specification (J276) would not have helped. That specification was designed to run with a practical programming project worth 20% therefore paper 2 looked somewhat different, without the in-depth scenario-based “Section B”. Each programming question on J276 was fairly short and self-contained, so didn’t demand a great deal of computational thinking. Add to this the COVID-adjustments to boundaries last year, against the first run of J277, and some schools may have gained false confidence in their performance.
Let’s look at some of the specific complaints from the social feeds, and I’ll try to refute them with evidence from the exam board as necessary. I’ve taken comments from the J277 and older OCR Comp Sci group (which I call the J276 group below).
What was said on the socials…
I can’t agree with these comments:
- “Bring back teacher assessed grades” and “I strongly believe teacher assessed grades should be implemented after this tragedy” – please, just no. Read this.
- “I’m going to email my kids tomorrow to let them know we all feel the same about the paper” and “Please encourage pupils and their parents to write/complain to OCR and other bodies such as OFQUAL etc. Do everything you can!” Please don’t. This will increase anxiety and be counter-productive, and possibly bring your school or the exam board into disrepute, so I’d advise against this.
- “People creating these papers need to show their face and take responsibility. The paper today seriously undermines teachers across the nation of this difficult subject, and can’t imagine what pupils are going through right now. If we don’t make noise, students in the future will continue experience days like this.” OCR have a feedback form, an appeals process and a helpful subject advisor, I strongly urge you to use the appropriate channels and take guidance from your exams officer or SLT before taking any other action.
- “most of us are 1 person subjects and the consequences from this are we have to explain ourselves to SLT and it makes us question our own ability. Almost feels like they want this subject to fail.” The correct response from SLT to any 1-person departments struggling to get good results in this subject would be to support you with CPD or recruitment. I’m aware this is difficult in the current climate, but all subject leaders have a responsibility to communicate upwards effectively about their department’s strengths and weaknesses. It goes with the territory, although I’ve been there and I know how hard it is. I give lots of actionable advice below.
- “loopy loop question driving me loopy let alone my poor EAL students!” – I thought the wording of 3b was about as clear as it could be, while still asking an important question that reveals whether the candidate understands the two types of loops in the context of a sorting algorithm. I talk more about this below. As for EAL, some students are allowed translation dictionaries under specific rules, but it is an English exam board GCSE.
- “It certainly wasn’t written by a teacher, as good teachers know how to relay computing concepts to students with a range of abilities.” – I think there were a lot of AO1 marks in Section A that are “easy” if the content has been taught well, retrieved often and revised well. Differences between HLL/LLL, arithmetic operators, define syntax and logic errors, spot the truth tables, describe features of an IDE to name a few. There is enough there to be accessible by those with mock/predicted grades of 1,2,3,4.
The accessibility of the paper is questioned several times on the Facebook groups. I think it’s worth exploring this some more in the context of the 3b insertion sort question. I put the question into Word and it came out as “Grade 6.7” so roughly UK Year 8 reading level. I think any attempts to dumb down this question would fail, because ironically the first sentence is an important preamble, conveying content rather than asking a question. The extra words are intended to help, not hinder understanding.
The answer “because the inner loop moves the unsorted number leftwards in the array and only stops when the number to its left is smaller than it, which is a condition not a count.” could only be gleaned from this question worded as it is, or similarly to it.
Amusingly I posted the question into ChatGPT followed by the prompt “Please help me reword this exam question to be easier to understand” and the result was as follows:
Why is it necessary for the inner loop in an insertion sort algorithm to be condition-controlled rather than count-controlled?
Note that we’ve lost the preamble, so this version could well be harder for some to answer than the original! We’ve lost the word “loop” after both “condition-controlled” and “count-controlled”, but does this make it easier or harder to understand? Ironically, Word now rates this “Grade 12.9” or Year 13/14 reading level due to the average word-length and sentence-length increase. I think we do examiners a disservice when we jump to conclusions about readability: question writing is not an intuitive skill.
The Logic Gate question was not as heavily scaffolded as last year’s question. But it was fair. Students need to understand that a boolean value often represents a fact about the world, such as whether it is day or night. It is reasonable to expect them to work out the logic circuit for an alarm system. I always teach logic circuits with real-world examples, because it says in the specification, “Understanding of how to create, complete or edit logic diagrams and truth tables for given scenarios”, and there was even a scenario question on Practice Set 2 Paper 2: “A cinema uses the following criteria to decide if a customer is allowed to see a film that has a 15 rating…”
Indeed, one of the most popular commercial resource bundles for our subject, that from Paul Long includes an exercise that is almost identical to last Thursday’s exam question, and the helpful advice “You could be presented with a real-world scenario and asked to create a logic circuit for that scenario”:
I do consider these valid complaints about the paper:
- Page 13 uses “alarm has been activated” when it should say “system has been armed” to match the variable name above “SystemArmed”, and to better describe the condition of being armed. Activation should only refer to the triggering of a sensor, otherwise the candidates will be confused. So the sentence below the bulleted variable list should begin “The alarm will only sound when the system has been armed…”
- There is an error in the identifier of the array on page 17, this should have said arrayEvents[1, 1] not events[1, 1]
- Printing the array on a right-hand page, with the algorithm writing space on its reverse caused unnecessary back-and-forth. Papers are usually designed to avoid this but not in this case.
The complaints about “Do Until” in question 1d are misguided. Both switch/case and do-until are in the specification, and switch/case was even on the 2022 paper. The Examiners’ Report (available on Interchange since last September) says this: “Candidates appeared to struggle with this question. In particular the use of switch/case was not well understood. This may be because some high-level languages such as Python have not traditionally supported this.” It’s important to teach the whole specification, and remember this is not a Python exam.
So the subject leaders need to know the specification inside out, read the examiners comments, attend OCR training and generally be experts in what the qualification is testing. As well as this, all teachers of the subject must have clear guidance on what to teach, be supported with quality materials, and most importantly, be provided with quality CPD so they can improve their subject knowledge. I’ve written before about the need for computing teachers to upskill themselves, so they can teach the subject better. In my June 2021 blog post, I said this:
Once you know it yourself, and feel confident you know it, you can explain the material in ways others understand. Rather than asking for slides and worksheets, I recommend teaching yourself the content. Then study others explaining it well.“Never Mind the Powerpoint“, June 2021, this blog.
Next steps for subject leaders
If you are a subject leader of computing…
- Don’t add fuel to the fire. You have a duty of care to remain calm, supportive and professional. Help your students by not exaggerating the issue and don’t encourage them to complain.
- Don’t assume we are all of the same opinion. As a subject body, opinions are at best divided on the quality and fairness of J277/02 2023. We are not universally outraged, perhaps step out of the “Facebook filter bubble” and see other opinions. I have made the case above that it is hard but fair, and many others share my opinion.
- Thoughtfully consider how you feel about both papers (and remember there was no great panic over paper 1) and give your feedback to OCR via the feedback form.
- Discuss with your exams officer what happened, and get a meeting with your SLT link this summer to start conversations about closing any gaps in you delivery, what help do you need?
- Request copies of completed scripts in August, with permission from the students, of the top, middle and bottom of your cohort. Read these alongside the examiners report, that comes out in early September. Use this to identify gaps in your delivery that may need closing with CPD, resources or curriculum changes.
- Use the NCCE and CAS and the free resources available online to upskill yourself and other teachers in the department. For example “switch/case” and “do/until” are explained in this Craig’n’Dave video. Book you and your department onto some subject-specific CPD, and buy books you can use to upskill (and yes I wrote one that is well-regarded, see the home page).
- Think broader than KS4, are results this year likely to be poor because of insufficient curriculum time at KS3, or non-specialist provision? Write those things on your subject improvement plan. We cannot be expected to deliver strong results with one hand tied behind our backs.
- Complete the Computing Quality Framework questionnaire which identifies the needs of your department. Use this to justify to SLT any support or changes you need: computingqualityframework.org
- Read the Ofsted Research Review of computing identify any gaps, and put these on your plan as well. Be honest and professional. (Oh and remember as computing teachers we are like hen’s teeth right now, if your current SLT don’t support you, you have options!
Computer Science is an EBacc subject and highly prized by employers and colleges. Ofsted’s recent report (above) makes it clear that they are expecting you to offer the GCSE (and also deliver alternative computing teaching at KS4 for those that don’t take it), and also that a minimum of 1 hour per week at KS3 delivered by a specialist is expected. Please resource your computing department accordingly and be guided by the subject leader in what they need to succeed.
To students reading this: remember you sat two papers, this was a tough paper but your scores will be aggregated, then standardised and grade boundaries will be set that reflect the difficulty of the papers. You have likely done better than you think. Put it behind you and do the best you can when exams resume on Monday. You’ve got this.