When you put up shelves, the first stage is to fix brackets firmly to the wall. After watching you carefully drilling and screwing, an observer might think you were mad if they then saw you try to pull them down again. But you know that if the shelves are going to support anything heavy, then they need to be very secure. You don't want it to come down, but you still try your hardest to make it fail.
This rigorous kind of attempt to find weaknesses should be what testing is all about in a learning context. You want to find the flaws, the misconceptions, the places where the structure might fail. I think this is true of high-stakes, end-of-the-road summative assessments or more frequent formative assessments, including informal conversations in class.
It is often said that testing must be positive: we're not trying to catch you out; we simply want you to show what you can do. But that is not testing. Inviting the shelf to choose its own load, it may proudly declare itself able to support a feather when it is capable of a great deal more.
You only find out its limits by testing to destruction.
Multiple-choice questions can be a good method, provided they are not constructed simply to be invitations to agree with a well-worded statement.
We should, in the nicest possible way, be trying to catch people out. Not in an aggressive, hostile way, but as a route to strengthening them for the future. "Trick" questions should not be seen as illegitimate. "Negative marking", in which marks are deducted for wrong answers, need not be viewed as something negative but as a realistic way of rewarding areas in which your ability is strong and penalising you for areas where it is weak. I rather like the idea of exams in which you answer questions and alongside indicate how confident you are of your answer, with the marks (positive or negative) being multiplied by your "confidence rating" - a bit of a gamble at times, maybe, but a great way to encourage learners to be honest and self-aware. The culture of "it's OK to guess - it's better than saying you don't know" seems to me a very unhelpful one. The schooling and exams process encourages it, but I don't want my doctor to work in that way - which, presumably, is why medicine is one of the few areas to adopt negative marking. In fact, I don't want the guess-and-try approach for anything that really matters. It is much better sometimes to say "I don't know".
To help learners to develop robust knowledge, skills and understanding, we have to be bold enough to test properly. If we fear too much the effect on learners' self-esteem of a collapsing structure, we set them up to fail in future, when it may matter much more. We need to encourage a view that you are more important than what you know or can do, that it's OK to fail from time to time. Being realistic about what you can and can't do today helps you progress tomorrow.
Colin Foster teaches at a secondary school in Coventry