Decades ago, group work meant hiding from the class clowns as they tried to turn me into Pinhead from Hellraiser with a hot-wire plastic cutter. We had been told to build a mosque from polystyrene. Fast forward to now, to me, post-trauma, actually in charge of a class, and I’m not saying you’d have to put a gun to my head to make me set group work, but I’d have to know you were packing heat.

Across the world, children are enthusiastically learning about Tudor belfries in trios and quartets, while flaying sugar paper and gobbling coloured pens and glue sticks. Group work: I hate the concept, as I hate hell, all Montagues and thee. I bite my thumb at it.

All my career, I have struggled with this contradiction. Every book I’ve read and trainer I’ve listened to have said it was the best way to learn, but I would set tasks and watch as, predictable as a metronome, the following would happen.

1. Uneven loading, as one or two children inevitably did most of the work.

2. Withdrawal, as the laziest fell into roles that suited their task-dodging talents.

3. Time-wasting, as the capacities of the many were engaged in the flabbiest way.

4. Unfair outcomes, as some groups produced good work from gifted key players and others didn’t.

I saw this time and time again, even if I had prepped the classes with rules for group work, defined their roles clearly and assigned groups in balanced ways. I saw it even though I persevered - and, I promise you, I am reasonably handy in the classroom. It was always a patchwork quilt of outcomes: some would enjoy it and engage, others would throw it off in disgust. And, like a break-up, I realised: it wasn’t me, it was you, group work. You’re a fine thing to leaven the tone and pace in a series of activities. In some circumstances, you’re Essential Football or cheerleading pyramids - but an efficient way to learn? Not so much.

The research that supports the adoption of group work is, as far as I can see, based on tiny samples or speculative interpretations of data. Usually such research concludes with: “Teachers need to be trained to use group work in order for it to be effective.” In other words, if it didn’t work, you weren’t doing it right. But teachers work in the real world and, human nature being what it is, if you allow some students to pause and lean, they will. That’s not an annoying glitch in the data, that’s the parameter of the intervention.

Fans of cooperative learning will insist that students need to learn the social skills of working in teams. I see no reason to believe that children won’t learn it regardless, simply by interacting normally. Such things are value claims rather than factual ones.

Using group work, it might take students half an hour to find something out that you could tell them in a few seconds. Even if the understanding stuck a lifetime (and I say it wouldn’t), then that’s a loooooong way to learn. Anyone got a spare decade to teach a syllabus?

Tom Bennett is a teacher in East London and director of the ResearchED conference