gap. This is important since on average the home learning environment scores of boys were
somewhat lower than those of girls, suggesting possible gender differences in parenting
practices.
Type of provision
Type of provision was not significantly related to attainment at entry to pre-school, when account
was taken of differences in intake in terms of child, parent and home environment characteristics,
but differences in cognitive progress related to type of provision emerged during the pre-school
period.
The multilevel analyses controlled for differences in duration of pre-school, as well as child
mobility (change of centre), since these were significant influences on cognitive progress for
several outcomes. In addition, as well as individual child, parent and home environment
measures, the analyses took account of compositional influences. A compositional influence is
associated with the cumulative effect of the social backgrounds of all the children in each pre-
school centre. Some centres may draw children only from a socially deprived catchment area.
The compositional effect in these centres would be different from a catchment area that
maintains a better social mix of children. Children in centres that served a higher proportion of
children with highly qualified mothers (had a degree/higher degree or professional qualification)
tended to make more progress in some outcomes, particularly pre-reading. Private day nurseries
tended to serve more children from educationally advantaged backgrounds. If compositional
effects are not included in the model, this form of provision showed significant positive effects for
pre-reading progress. The inclusion of controls for child mobility, compositional effects and pre-
school duration mean that the extent of differences between types of provision identified by the
model is reduced (because such factors are themselves related to type of provision).
Outlier centres, both positive and negative, were found in each type of provision. There was
significant variation in effectiveness on cognitive progress within each type of provision; thus we
can conclude that differences between individual centres are likely to be more important than
differences between type. Nonetheless, certain patterns emerged suggesting that some forms of
provision were generally more effective. Integrated provision (i.e. combined centres) showed a
significant positive impact for several measures. Nursery schools also showed some positive
effects compared with other types of provision similar to those found for integrated provision. By
contrast, children who attended local authority day nurseries tended to make relatively poorer
progress, especially for pre-reading. There were interactions for low SES children with type of
provision. Children in the low SES group showed better outcomes if they attended an integrated
provision (i.e. combined centres) or nursery schools. Both these forms of provision also showed
higher scores in observed quality.
While private day nurseries did not show up as significantly more effective in the analyses of
impact of type of provision (except in comparisons with local authority day nurseries for pre-
reading and language), a number of positive outlier centres for pre-reading were found to be
private day nurseries. The results suggest that centres classified as private day nurseries in
particular show much variation in effects and quality, some having a specific educational
philosophy or tradition (e.g. Montessori). Each centre was categorised by the percent of
children’s mothers with university education. This percent was related to the amount of cognitive
progress made by children in the centre. A ‘compositional effect’ such as this suggests that
concentrations of disadvantaged children (in terms of mothers’ educational levels) is related to
lower developmental progress in children. This suggests that the clustering of disadvantaged
children within specific centres is not conducive to their cognitive progress. Policies aimed at
encouraging a social mix of children may be more appropriate, although this may be difficult to
achieve in practice, given many parents' preferences for centres close to home, and the extent
that social and ethnic groups cluster in some neighbourhoods.
There was significant variation both between individual centres and by type of provision in the
observed quality of provision (see Section 5). When account is taken of variation in quality of
centre environments, the impact of type of provision is reduced. This indicates that the impact of
type of provision is likely to be, at least in part, attributed to variations in environmental quality
28