We’ve seen a lot of problems with exams recently – just look at the problems last summer with mistakes in a wide range of exam papers. Today I’ve found that AQA have spent so little time checking that suitable research sources are online that the only good Google results are their own teacher notes, and a primary science investigative cartoon. On top of this, a new specification inevitably means a lack of practice material which means students and teachers don’t really know what to expect. If you have to explain why this is unfair to non-teachers, perhaps this analogy might help; we wouldn’t expect to have a driving test on the road having only practised in car parks, would we?

I have an idea.

In fact, I have *two* ideas, neither of which is mine. If we take the ‘backward design’ principle (originated by Wiggins and McTighe, introduced to me by Robin Millar’s work) and combine it with a ‘curated crowdsourced’ model, maybe there’s a way to do a better job.

**Backward Design**

My apologies to Robin and other experts if I miss the subtleties – I’m just a classroom teacher with delusions of writing grandeur. Instead of beginning a syllabus with the content that we want to teach, backward design asks what we want students to be able to do at the end – how will they be tested? How will we know if the course was successful or not (or more precisely, how successfully the student has completed it)? If we create **assessment tasks that will allow us to differentiate between students** – ideally including, but not limited to written exams – then we can develop a list of what students should learn, which gives us a list of possible learning/teaching activities. As Robin and others point out, ‘teaching to the test’ is only a problem if the test is not fit for purpose. If we produce a realistic, useful test then being prepared for it is a positive thing.

**Crowdsourcing**

So who better to contribute possible questions than teachers? Imagine a Google form set up by a new exam board; let’s call it CCEB. Exemplar material, based on accepted good practice, shows how to lay out mathematical working. Questions are entered, with a markscheme. Dropdown boxes allow those entering the question to define marks available, and from key words describing the area(s) of science being tested. Active teachers, retired staff, academics – even students – all can contribute. The contributions are freely given on the basis that the results will be freely available as far as practical, probably via Creative Commons licensing.

**Curated**

When a certain threshold is reached – which if every science teacher in the UK supplies a single question, won’t take long – the submissions are sorted by category and checked by CCEB staff. Because they are being proofread rather than written, it will be quicker and easier. If you have some of the original contributors – determined by random allocation – paid for a day’s work, they can be pre-moderated as well. Mathematical questions can be kept in the same form but with different numbers substituted. A large pool of questions is now complete, ready for the exam, which can be balanced between topics. There will be enough questions, all produced at the same time, for several specimen papers to be made available. With a large enough pool, you could even make all the questions open source, like those for the theory element of the UK driving test.

**One Day**

It’s feasible that in the future, with enough questions available, every student could get a different but equivalent exam, as described in John Barnes’ book Orbital Resonance.

In the meantime, maybe we as science educators can get involved with setting better exams than the ones we complain about. The exam boards could ask for submissions in this way now. The cynic in me thinks that this would make it much harder for them to justify their existence. Maybe they would like to prove me wrong.