A teacher’s belief about a pupil does two jobs at once. It predicts how the pupil will do, and it can change how the pupil does. The second job works as a prophecy that fulfils itself in the classroom.

The mechanism runs through the pupil. What a teacher expects can shape how a child rates their own ability, and that self-judgement feeds into behaviour and marks.

The effect can reach well past school. Nicholas Papageorge and colleagues put a number on it in 2020. Their model gave the college graduation rate an elasticity of 0.12 against what teachers expected. Put plainly, a higher forecast went with a higher chance of finishing university.

How close is the forecast to the result?

Close, most of the time. Researchers writing in Scientific Reports compared teacher judgements with national exam results at ages 11 and 14 in a UK cohort study. They found the two strongly correlated.

The misses were not random, though. How accurate a teacher’s forecast proved depended on the pupil’s socioeconomic background. It did not depend on anything the researchers measured about the teacher.

The pupils teachers mark down

The same paper summarises earlier work on which pupils teachers tend to mark down in their heads. On average the list covers four groups:

  • pupils with special educational needs
  • pupils of black-African and black-Caribbean origin
  • pupils from poorer homes
  • boys

One meta-analysis the authors cite pooled 39 studies. It found teachers held higher expectations for European-American pupils than for pupils from ethnic minorities.

A low forecast has consequences beyond a single grade. The authors note that a teacher’s view can decide which subjects a pupil takes and whether they join an advanced stream. It can also set the level of exam they sit and how long they stay in education.

An inflated forecast carries its own risk, the Scientific Reports authors add. A pupil placed in a stream above their level may not get the support they need to reach their potential.

Why a teacher’s view still counts for something

Teacher judgement has real strengths over a single test. It rests on months of watching a pupil, so one good or bad exam day does not distort it. It also removes the pull to teach to the test.

The UK leaned on it heavily during the Covid-19 pandemic. Teacher assessments decided results after the government’s first statistical model for exam grades drew heavy criticism.

Retraining the teachers with low hopes

One recent study tried. XinJuan Li and colleagues worked with 115 teachers and 356 pupils in the third grade, spread across five primary schools. Teachers judged to have low expectations went through a 12-week training programme built on cognitive behavioural therapy. Teachers with high expectations served as the comparison group.

The trained teachers’ expectation scores rose, with an effect size the authors put at a Cohen’s d of 0.72. Pupils were tested in Chinese and maths. Across five rounds of testing, the share scoring above 80 points climbed from 40.4% at the start to 76.2% at the end.

Behaviour moved too. Pupils taught by the trained teachers showed more positive behaviour and less negative behaviour than at baseline.

The design has a limit worth knowing. It was quasi-experimental, with a control group that was not matched to the trained group. That makes the rise in scores suggestive rather than proof of cause.

What the genes added

The Scientific Reports team also asked whether genetics played a part. Forecast accuracy was linked to a pupil’s polygenic score for educational attainment. They found no strong evidence that teacher accuracy itself was heritable across the genome.

For a parent, the practical question is what the teacher expects, and whether anyone has checked it against the child’s work.