Homework, an effect that depends on age

A 1956 circular bans written homework in primary school, and that ban has been restated several times since, which indicates it is not applied. The ban covers written work; lessons to learn and reading fall outside it, which leaves a zone of interpretation that practice has largely occupied.

The grounds invoked at the time combined the length of the school day and inequality in working conditions at home, two arguments still present in today's debate.

Harris Cooper and colleagues published several syntheses of American research. They find an association with school results that grows with age: close to zero in primary school, moderate in lower secondary, clearer in upper secondary. Their work, published from the late 1980s and updated in 2006, is among the most widely cited in the international literature.

Those results are mainly correlational. Pupils who do more homework differ from the others, and the amount of homework set also depends on the level of the class.

Several studies find a non-linear relationship: beyond a certain daily duration, the association stops rising and may reverse.

That result is cited by supporters of capping the amount. Critics stress that a struggling pupil takes longer for the same work, which would produce the same curve without any negative effect of volume.

Both readings are compatible with the same cross-sectional data. Only arrangements that vary the time required independently of pupils' level would make it possible to decide between them, and they are rare.

French sociological work, notably that of Stéphane Bonnéry on school work outside the classroom, documents gaps in material conditions, in available help and in understanding what is expected. Those gaps are not only a matter of resources: they also reflect familiarity with the school's implicit expectations, which some families share and others do not.

The argument is not about the average effectiveness of homework but about its differentiated effect: the same instruction does not produce the same work depending on the home, which moves part of the attainment gap outside school.

Almost all the work measures quantities, not content. The type of task, its novelty, the feedback the teacher gives are rarely documented, although those are the variables teachers actually manipulate.

The few trials varying the type of task rather than the volume are small and do not yet allow a general conclusion. Public debate, for its part, continues to focus on whether there is homework or not, that is, on the least informative variable.