Schooling at two, a policy in retreat

Schooling before three peaked in the late 1990s then fell continuously, linked both to staffing decisions and to the evolution of early childhood provision. France was then an exception, as few European countries took in children of that age in a properly school-based setting.

A revival scheme was launched from 2013, targeted at disadvantaged areas, with results below the stated objectives and strong departmental heterogeneity.

Work published by the ministry's statistical services and by researchers finds weak, even null average effects on attainment measured later.

Some studies find positive effects concentrated on children from disadvantaged or non-French-speaking families, which echoes the international finding about the role of the available alternative.

The difficulty also lies in measurement: attainment assessed several years later depends on everything that happened in between, and a real but transient effect may vanish from the measures, as has been observed for other early childhood programmes.

Criticisms bear less on the principle than on the conditions: class sizes, teacher training for that age, adaptation of premises, coordination with nap time and care. Those conditions vary greatly from one school to another, which makes any general conclusion drawn from a particular school hazardous.

Defenders of the scheme argue that a suitably adapted school is preferable to no provision at all, and that the debate about conditions should not be confused with a debate about whether to do it. The two positions agree on one point, that the effect depends on what is actually offered to the child, and diverge on what that implies for the scheme.

Children schooled at two are not randomly assigned: they live in areas where provision exists and in families that request it. Comparing their results with others mixes the effect of the scheme with that of the context.

The most solid analyses exploit variation in provision between areas and over time, and they remain few in France. They come up against the difficulty of obtaining comparable attainment measures over a long period, and the small scale of the usable variations in provision.

Several European countries have no school before three but have highly developed preschool provision, which makes comparing enrolment rates misleading. A two-year-old may thus spend their days in an educational setting without being counted as enrolled in school.

The relevant question is not the school status of provision but its concrete characteristics, which international statistics render poorly because they classify by institution rather than by content.