Training teachers who haven't taught yet

Preparing a teacher is harder to measure than developing one. An experienced teacher has a track record — results, a reputation, years of observed practice. A trainee has none of that. The programme has to judge readiness on a handful of supervised lessons, and much of that judgement lives in a mentor's head rather than in a record.

That creates three problems the Aga Khan Academy wanted to solve. Feedback varied by whoever happened to observe. Progress across the programme was described rather than demonstrated. And the trainees themselves had no clear picture of where they stood — which matters more for a novice than for anyone else, because a new teacher's biggest risk is not knowing what they don't know.

One rubric across the whole programme, and every observation of every trainee scored against it.

• Expectation-mapped criteria. Observers select the written statement describing what they actually saw, and the rating follows from the statement. A trainee reads why they were rated as they were, not just the number.
• Evidence attached in the room. Comments, lesson plans, board photographs and samples of student work sit against the specific criterion they support, so feedback points at something rather than describing it.
• Self-assessment before the observer's. Each trainee rated their own lesson on the same criteria before seeing anyone else's rating.
• Tasks between observations. A criterion below expectation generated specific work with an owner and a date — and the next observation re-checked that criterion first.
• Tracking across cohorts. Both cohorts sat on the same instrument, so cohort two could be read against cohort one rather than assessed in isolation.

For trainee teachers, the gap between how they rate themselves and how an observer rates them turned out to be the single most useful measurement in the programme.

A trainee who rates themselves well above their observer has a calibration problem — they cannot yet see what a strong lesson looks like, so they will not know when to change course. A trainee who rates themselves well below has a confidence problem — often teaching better than they believe, and at risk of leaving the profession early for the wrong reason.

These need opposite conversations, and neither is visible from an observer's rating alone. Running self and observer assessment on the same criteria makes the gap a number a mentor can act on.

Running both cohorts on one instrument turned the second into something more useful than a repeat of the first.

An experienced teacher can be told what to fix. A trainee has to learn to see it themselves — which is why the self-rating is not a formality on this programme.

How the instrument was designed for pre-service

• IMPACT works pre-service, not only in-service. The same architecture that measures a practising teacher measures a trainee, which means a school can carry one record from training into employment.
• Cohort tracking makes a programme testable. Two intakes on one instrument turn programme design from intuition into something with evidence behind it.
• Small cohorts still benefit. Twenty teachers is not a statistical sample, and we do not report it as one. What it produces is consistency of judgement and a complete record per trainee — both of which matter more than significance at this scale.