Several research projects have indicated that the nature of the neighbourhood in which a child lives may affect development over and above the influences of the child’s family. For example in a population representative longitudinal study in the USA it has been found that living in more affluent neighbourhoods is associated with better child IQ scores over and above effects of family socio-economic status (SES) and parent education (Chase-Lansdale, Gordon, Brooks-Gunn, & Klebanov, 1997). Some have claimed that such neighbourhood influences may explain five to ten per cent of the variation in academic attainment (Leventhal & Brooks-Gunn, 2000). Such neighbourhood influences would include the perceived value of education in that neighbourhood, schools’ and children’s access to resources from the neighbourhood, e.g. libraries, parks and recreational facilities, and also community-based efforts to support child development.

Much research concerning the influence of ‘neighbourhood’ effects has focussed on adolescents (e. g. Leventhal & Brooks-Gunn, 2000; Sampson, Raudenbush and Earls, 1997). While

adolescents may well experience greater exposure to neighbourhood features such as peer groups, ‘role models’ and general milieu, the neighbourhood as a source of influence may also be important for younger children. In the USA, the Committee on Integrating the Science of Early Childhood Development (2004) suggested that the interaction between pre-school children and their relatives, neighbours, religious communities, child care and health systems may well lead to neighbourhood influences beginning long before adolescence. Thus there is reason to consider neighbourhood influences on young children’s development and behaviour.

Some existing evidence indicates some small effects for young children’s development associated with the neighbourhood. In the USA, Chase-Lansdale, Gordon, Brooks-Gunn,

Klebanov (1997) found around two per cent of the variation in behaviour problems and academic achievement for 5 and 6 year olds was linked to neighbourhood effects (deprivation and ethnic diversity). Similarly in the UK McCulloch and Joshi (2001) found 4-5 year olds achieved lower cognitive scores if they came from poorer rather than more affluent neighbourhoods

independently of other socio-economic measures. Also in the analysis of data for over 500,000 children per year for three successive years (2002-2004) in all state primary schools in England, Melhuish et al., (2006a; 2006b) found that children’s progress from Key Stage 1 (age 7) to Key Stage 2 (age 11) was also influenced to a small extent by the level of deprivation of their neighbourhood. However, it is possible that such ‘neighbourhood’ effects may reflect

unmeasured differences in families resulting from the non-random distribution of families across neighbourhoods.

All research discussed so far deals with the issue of neighbourhood effects by seeing whether there is a separate influence associated with neighbourhood deprivation after standard child and family demographic factors, such as child gender, ethnicity and age, and parental socio-

economic status (SES) and education, have been taken into account. Such research does not include data on families as rich as that in the Effective Pre-school and Primary Education 3-11 Project (EPPE 3-11) research. Thus it is possible with the EPPE 3-11 data to investigate neighbourhood influences including more control of child and family factors than has previously been achieved. In particular the EPPE research has developed a measure of the learning opportunities provided within the home, the Early years home learning environment (HLE) index and this measure has proved to be a powerful predictor of educational achievement (e.g.

Melhuish et al., 2008a) and social/behavioural development (e.g. Sammons et al., 2008c).

Measures of the neighbourhood

Other EPPE 3-11 reports concerning the basic influences on children’s cognitive performance (e.g. Sammons et al., 2007a) show the influence of child, family and home learning environment characteristics on attainment and progress. In addition to these factors three measures of the neighbourhood are considered: two measures reflecting the parents’ perceptions of their neighbourhood in terms of social cohesion and safety, and the Index of Multiple Deprivation (IMD).

The neighbourhood perception measures were derived from a parent survey completed when the EPPE 3-11 child was approximately six years old. The responses to questions related to the neighbourhood were entered into a principal components analysis (varimax rotation) and two clear factors emerged. These neighbourhood factors are described below in terms of the items that loaded most heavily upon the factor.

For one neighbourhood dimension, there were four questions concerned with neighbourhood safety that loaded on the same neighbourhood factor, and were related to the perceived frequency of: violence or crime involving people; violence or crime involving property; general nuisance; and sense of safety when walking alone after dark. These items showed good inter- item consistency (Cronbach’s alpha = 0.75) and appeared to provide a coherent neighbourhood dimension.

For the second neighbourhood dimension, three questions loaded on the same neighbourhood factor, and were concerned with social cohesion and involved the perceived frequency of neighbours: doing favours for each other; sharing information on schools or children’s activities; visiting each other’s houses. These items also showed good inter-item consistency (Cronbach’s alpha = 0.84) and again appeared to provide a coherent neighbourhood dimension.

In addition to the above measures of perceived safety and social cohesion, the IMD score was used as a neighbourhood measure. The IMD is a nationwide index combining weighted measures or levels of: crime; barriers to housing; living environment; education and skills training; health deprivation and disability; employment and income (for further details of the IMD see The English Indices of Deprivation 2004: Summary [revised], 2007). Greater IMD scores indicate greater levels of area deprivation. The 2004 IMD scores were assigned to each child on the basis of their postcode.

The three neighbourhood measures showed only small associations with each other indicating that they were relatively separate dimensions of the neighbourhood (safety - social cohesion rho

= .09; IMD - safety rho = -.32; IMD - social cohesion rho = -.15;).

The neighbourhood measures were moderately associated with family demographic measures. Greater neighbourhood safety and social cohesion and lower deprivation (IMD score) were associated with greater family social advantage (defined as a combination of parental educational, occupational and income status). Similarly there was a moderate association between the neighbourhood measures and the Early years home learning environment (HLE) measure with families in more advantaged neighbourhood tending to score higher on the Early years HLE index.

Does neighbourhood have a separate effect upon educational achievement?

Firstly we considered children’s ability in Reading and Mathematics as measured by

standardised assessments at age 6 years. These outcomes were analysed firstly in terms of the standard child and family demographic variables, then the neighbourhood variables were added to the analysis to see if they showed an additional effect, and finally the Early years HLE

measure was added to see if neighbourhood effects were altered when the Early years HLE was included.

With Reading there were no significant effects associated with any of the neighbourhood variables. With Mathematics, the IMD score was associated with a significant effect that

persisted even when the Early years HLE measure was added into the analysis. The IMD score had a small significant additional effect (ES=0.13) on Mathematics scores at 6 years of age, whereby children in areas of higher deprivation scored lower even after taking account of all child, family, and Early years HLE effects.

This strategy for investigating the possible effects of neighbourhood variables was repeated when children were aged 11 years. At this age the children’s performance in English and Mathematics in the Key Stage 2 National Assessments were used as the measures of

educational achievement. When only child and family demographics are included in the analysis, the addition of the IMD score was associated with a significant additional effect. However when the Early years HLE measure was also added the effect for IMD became insignificant. This pattern of results was reflected for both English and Mathematics whether considering either attainment or progress from age 7 (Key Stage 1). It is possible that inter-family differences may mediate neighbourhood effects. Family characteristics and neighbourhood characteristics can co-vary, and when examined together family characteristics tend to supersede or displace the neighbourhood measure but without one being reducible to the other. This may well be happening. When we include the Early years HLE variable in analyses the effects for

neighbourhood become insignificant, yet without the Early years HLE in the analysis there is still a significant neighbourhood (IMD) effect. Such an interpretation suggests that the process by which the neighbourhood influences pupil outcomes is mediated through effects upon within family processes that give rise to different family environments, reflected in changes in the HLE, which in turn affect child development and educational achievement. For example decisions concerning activities with children may be influenced by the wider peer group, i.e. peer group learning may be operating amongst parents.

Does neighbourhood have a separate effect upon social/behavioural development?

Controlling for child, family and Early years HLE characteristics, none of the neighbourhood measures had statistically significant effects on any of the four social/behavioural measures at Year 6. Considering that in these models we controlled for specific family level characteristics, it is likely that these specific family level predictors suppressed any neighbourhood effects. Similar findings were evident for the sample of children from all schools in England where census-

derived data had a stronger effect on pupils’ educational outcomes and therefore suppressed any IMD effects (see also Melhuish et al., 2006a; 2006b).

In document Final Report from the Primary phase: pre school, school and family influences on children’s development during Key Stage 2 (7 11) (Page 111-114)