Small disruptions in a classroom

Why does a teacher who stops to address trivial interruptions end up losing less lesson time than one who ignores them?

A pupil calls out an answer without being asked. It takes two seconds. Stopping to deal with it takes fifteen. On that arithmetic, ignoring it is obviously cheaper, and a teacher who chases every small thing looks like someone spending a fortune to save pennies.

Yet the experienced advice runs the other way, and teachers who follow it seem to reach the end of the hour further into the lesson. Something in the two-seconds-versus-fifteen calculation must be wrong. Which term is it — the cost of an interruption, or the number of them?

Take the cost first. What does an interruption actually consume? Not the two seconds it lasts. Thirty people looked up, the sentence has to be re-established, and each pupil has to find their place again. The event is short; the recovery is not, and it is paid by everyone at once rather than by the interrupter.

So the unit cost is larger than it looks, but notice that this argument does not favour intervention — it inflates both sides of the ledger. If ignoring costs fifteen seconds of class attention rather than two, and addressing costs twenty-five, addressing is still the more expensive move. Per event, ignoring always wins. Which tells us the answer cannot lie in the cost term at all.

Then it must lie in the count. Here is the question that unlocks it: is the number of interruptions in an hour a fixed quantity, or does it depend on what happened to the earlier ones?

Think about it from the pupil's side. Whether calling out is worth doing depends on what the pupil expects to follow. Nobody runs that calculation explicitly, but everyone runs it — and the evidence they run it on is the record of what has just happened in this room, to other people, in the last ten minutes. An ignored interruption is not a neutral event. It is a demonstration, to thirty witnesses, that the thing carries no consequence. The rate is not an input to the lesson. It is an output of the lesson's own history.

Now the two policies stop being comparable per event, because they are not facing the same number of events. Put illustrative figures on it — these are a model, not a measurement. Say interruptions arrive at four in the first ten minutes. Ignored, the rate climbs by two each subsequent block: 4, 6, 8, 10, 12, 14, which is fifty-four events across the hour at fifteen seconds each — thirteen and a half minutes gone. Addressed briefly, at twenty-five seconds each, the rate instead decays: 4, 3, 2, 1, 1, 1, which is twelve events, five minutes. The expensive policy costs a third as much, and it does so entirely because the count moved.

Notice what makes this work: the intervention is cheap relative to the rate change it buys, not relative to the event it answers. That immediately tells us when it stops working. A response so long it becomes an event in its own right — a lecture, a confrontation, an audience — raises the unit cost into the range where the arithmetic reverses, and worse, hands the interrupter exactly the attention that made the behaviour worth trying. So the claim is narrower than "deal with everything". It is that a brief, early, low-drama response is the one that changes the rate at a price worth paying.

Three refinements hold this together.

First, the mechanism is about information, not punishment. What lowers the rate is that pupils learn interrupting is observed. Kounin's classroom observations in 1970 gave this the name "withitness" — the teacher who signals awareness of what is happening behind them — and the point of the concept is that the signal, not the sanction, does the work.

Second, the argument does not say interruptions are the main loss in a lesson. Transitions — starting, moving between tasks, packing away — consume more time in most classrooms than misbehaviour does, and a teacher optimising only for disruption is fixing the smaller line item.

Third, and least comfortable: the evidence here is weaker than the confidence with which it is taught. The classroom-management literature is largely observational — skilled teachers were watched, and their common practices catalogued — which cannot separate "doing this makes lessons orderly" from "orderly lessons let you do this". The famous urban version of the same idea, broken-windows policing, has not held up well under scrutiny, and it would be careless to lend its authority here. What does have repeated randomised support is the narrower finding that structured group-contingency programmes — the Good Behavior Game is the standard example — reduce disruptive behaviour in trials; I recall multi-site replications including long follow-ups, though I would not quote a magnitude.

So how would we know the mechanism is real rather than plausible? Everything rests on one claim: that the arrival rate rises when interruptions go unanswered. That is directly measurable. Count call-outs per ten-minute block across lessons, and compare blocks following an answered interruption with blocks following an ignored one. If the rate is flat — if a class produces its interruptions at a constant rate regardless of response — then the count term is fixed, the per-event arithmetic is the whole story, and ignoring is straightforwardly correct. That flat line is the observation that would refute this.

• Why transitions and lesson starts lose more time than misbehaviour, and what recovers them.
• How a group contingency changes who is enforcing the norm, and what that costs in other ways.

Every term the collection defines is gathered in the glossary.