Addressing Educational Disadvantage in Deprived Communities: Evidence from the New Deal for Communities Programme in England

Urban Regeneration Policy in England: The Context

Governments across developed economies have addressed problems in ‘pockets’ of deprivation in cities and large towns. This trend is apparent in schemes such as Empowerment Zones in the US (Oakley and Tsao, 2006) and European initiatives such as URBAN (Carpenter, 2006). Similarly, from the mid 1960s, Conservative and Labour governments instigated programmes to regenerate physical, social or economic conditions in English cities (Atkinson and Moon, 1994; Kintrea, 2007; Shaw and Robinson, 2010). ¹ These schemes, often referred to as area-based initiatives (ABIs), operated for specific periods of time, providing relatively limited resources to ‘deprived’ localities. Despite the proliferation of such schemes, by the late 1990s many deprived urban neighbourhoods remained. Partly in response to this, in 1998 the Social Exclusion Unit (SEU), outlined a rationale for a national strategy for neighbourhood renewal, the “most concerted attack on area deprivation this country has ever seen” (SEU, 1998, p. 12), one central component of which was the New Deal for Communities (NDC) Programme.

The NDC Programme, launched in 1998, was designed to “help turn around the poorest neighbourhoods” (DETR, 1998, p. 1), thus reducing “the gaps between … [these areas] and the rest of the country” (DETR, 2001, p. 2). The 39 NDC Partnerships were to attack problems within areas consisting of, on average, 9800 people. Ten were located in London and most of the others within deprived areas of city-regions including Birmingham, Manchester and Liverpool. Because each of England’s nine regions received at least two NDCs, some were sited in relatively less deprived cities including Plymouth and Norwich. However, a consistent pattern across all 39 was that they tended to be located in the most deprived areas within their parent local authority. Each Partnership was allocated a 10-year £50 million budget, the Programme as a whole costing around £2 billion. Each NDC was to achieve change by working with delivery agencies and by placing the local community at the heart of the initiative. Partnerships were charged with attempting to secure change across six outcomes. Three were designed to improve these 39 ‘places’: crime, the local community, and housing and the environment; and three outcomes were for local residents: health, worklessness and education.

There is a long tradition in English urban policy of addressing educational disadvantage within urban areas. More than 40 years ago, following the Plowden report, ‘Educational Priority Areas’ were launched to tackle educational disadvantage reinforced by social deprivation (Central Advisory Council for Education, 1967). However, the most insistent drive on improving educational standards within deprived areas emerged from the three Labour administrations elected between 1997 and 2010. A strong emphasis was then placed on driving a standards agenda supported by league tables and inspections by the Office for Standards in Education (Ofsted), which inspects and regulates services for children and young people. National literacy and numeracy agendas were also put in place, underpinned by a push towards higher-quality teaching. Various initiatives were adopted to weaken links between area-level deprivation and attainment. In all, 73 Education Action Zones, consisting of areas containing about 20 schools, were introduced between 1998 and 2000 (Ofsted, 2003). Similarly, the Excellence in Cities Programme, launched in 1999 in over 400 secondary schools, was intended to harness school-based interventions to tackle educational problems in deprived localities (Machin et al., 2003). More recently, school improvement programmes such as the London Challenge, 2003, the City Challenge Programme, 2008, and National Challenge, 2008, continued to focus investment in deprived localities.

Evidence began to emerge of improvements following increased investment during the 2000s (DCSF, 2009a). Nationally, average standards went up, with pupils and schools in more deprived areas making most progress. The proportion of children entitled to free school meals (roughly the poorest 15 per cent) obtaining five or more good GCSEs at Key Stage 4 (for 16-year-olds) including English and Maths increased from under 15 per cent in 2002 to 23.5 per cent in 2008. Similarly, the performance of the major census groups – Black, Asian and Mixed pupils – improved faster than the cohort average at both primary (under 11) and secondary (11–16) levels (DCSF, 2009a). However, these gains need to be set against the reality that at the end of the 20th century, the UK still had higher levels of inequality compared with other developed countries (OECD, 2001). In summarising evidence, Hirsch concluded that

children from disadvantaged backgrounds do worse than those from advantaged backgrounds by a greater amount than elsewhere (Hirsh, 2007, p. 3).

Differences in attainment between the children of affluent and less well-off parents continued to emerge at an early age (Carter-Wall and Whitfield, 2012; Goodman and Gregg, 2010). A 2009 UK government research report describes

a very clear pathway from childhood poverty to reduced employment opportunities … deprivation has a negative impact on educational attainment (DCSF, 2009b, p. 6).

Similarly, a 2010 review of poverty and life chances pointed to

overwhelming evidence that children’s life chances are most heavily predicated on their development in the first five years of life (Field, 2010, p. 5).

Nevertheless, schools can play a role in helping to enhance outcomes for deprived pupils (DCSF, 2009b; Lupton, 2006). Some schools are apparently more successful because of factors such as leadership, experienced staff bases and high quality teaching (Ofsted, 2008, 2009). Moreover, there is evidence suggesting that a poor physical environment can impact on children’s relationships with education (Hollingworth and Archer, 2010). This can create a situation whereby

disadvantaged neighbourhoods tend to have schools that are of lower quality than those in rich neighbourhoods (Lupton, 2006, p. 63).

This lower quality may be reflected in various ways

Pupils from deprived backgrounds typically have less access to a good, broad curriculum and related extension activities, and may find their curriculum irrelevant to their future … Teachers’ attitudes, assumptions and behaviours may be influenced by pupils’ socioeconomic background, and this may disadvantage pupils from deprived backgrounds (DCSF, 2009b, p. 67).

However, although schools play a role in explaining educational disadvantage, Field suggests that by school age, there are wide variations in children’s abilities; “schools do not effectively close that gap” (Field, 2010, p. 5). Taking a broad overview of evidence, school performance is a relatively minor contributor in explaining attainment: a 2007 study suggested that only 14 per cent of variation was attributable to school quality (Cassen and Kingdon, 2007).

Reflecting on a policy context in the 2000s wherein deprivation and spatial disadvantage figured prominently, it can therefore be argued that it was entirely appropriate for the NDC Programme to adopt educational outcomes, the narrative underpinning which is the focus of this paper.

However, a number of caveats should be flagged up here. First, this was never an ‘educational’ ABI, any more than it was a health or a crime initiative. The assumption was that the Programme would address an inter-related array of ills impacting on neighbourhoods. It would be unwise to compare results from this initiative with interventions more directly focusing on education, such as SureStart which has apparently shown beneficial impacts (DfE, 2010). Secondly, it is important to stress the community-based nature of this ABI (Wallace, 2007). Approaches adopted by NDCs were overseen by a Partnership Board. At some stage, local residents held a majority on all 39; in 2008, they did so in 26 (DCLG, 2009f). This is not the place to consider implications arising from this approach to governance, Yet it undoubtedly encouraged a kind of orthodoxy to perceiving problems and identifying solutions (Lawless, 2004). Crime was seen as revolving around anti-social behaviour, robbery and burglary and was best addressed by more police. Education was similarly generally perceived in a relatively unproblematic manner: attainment was the key issue and was best achieved through support to local schools. A community focus neither encourages innovation, nor the use of available evidence. And, thirdly, across all six outcomes theories of change were often non-existent, or ill-defined. Partnerships did not generally identify a set of evidence-based interventions which might plausibly change existing levels of disadvantage to desired outcomes over 10 years. Rather, projects were approved because of locally contingent factors: community demands, attitudes of agencies and available match-funding. It was widely assumed that projects would help to achieve outcomes, even if the processes through which this was to occur remained blurred. Consequently, it is not possible to trace the impact of specific interventions on particular outcomes.

This problem was accentuated because Partnerships had relative freedom in planning their strategies and interventions. A 2004 analysis of Delivery Plans established that, on average, each Partnership then assumed five separate educational outcomes, ranging from nine in Nottingham to one in Oldham in Greater Manchester. On average, each had funded 12 projects ranging from Hull with 33, to two in Southwark in south London (Marshall, 2005). More would have been learned from the Programme, if Partnerships had supported a narrower range of projects, each based on a theory of change linking levels of deprivation, interventions and plausible outcomes—but community-driven regeneration schemes do not work like that. What mattered was what local residents wanted and what agencies could deliver. Across all outcomes, not just education, the exigencies of ‘delivery’ dominated, not evidence-based reflection. However, one constant theme emerged in relation to education: the push on initiatives within schools to enhance attainment levels. By 2004, all 39 were funding projects to support schools and 34 were funding educational attainment schemes. And with regard to outcomes, 26 were hoping to improve Key Stage 4 attainment levels (for those aged 16). Remaining sections of this paper consider research methods, change for areas and residents, and the implications of findings.

The National Evaluation: 2001–10

Although this was an intensive ABI, the scheme embraced only 39 localities, representing only a tiny fraction of deprived English neighbourhoods. It was always the intention that an evaluation would be commissioned to identify generalisable lessons. To that end, the then Office of the Deputy Prime Minister (ODPM), later the Department of Communities and Local Government (DCLG), commissioned an evaluation, final reports from which were published in 2010 (for an overview, see DCLG, 2010a). The evaluation carried out a number of data analysis tasks, of which three are relevant to the narrative surrounding education: the household survey, administrative data and case study investigations.

First, four household surveys were carried out across all 39 areas. In 2002 a baseline was established using a survey questionnaire. This adopted a random sample design, involving the collation of 500 responses from all areas thus culminating in around 19 500 responses across the Programme in 2002 and again in 2004. For the 2006 and 2008 surveys, numbers in each area were reduced to 400 creating Programme-wide totals of about 15 800. The survey, which explored attitudes and change in relation to all six outcomes, was based on a combined panel and cross-sectional top-up design. Some 10 638 of 19 633 interviews in 2004 were held with the same respondents as in 2002. The same principle applied to later surveys. Due to attrition, it was only possible to revisit about 55 per cent of those responding two years previously. As a result randomly selected top-up interviews were held to maintain a sample in each location. Final evaluation findings with regard to panel evidence were based on 3554 respondents interviewed at all four waves (DCLG, 2010b).

The most significant problem in evaluating ABIs is that of the counterfactual: what would have happened if the initiative had not gone ahead? This issue is best addressed through benchmarks identifying change over and above that occurring elsewhere. Three benchmarks were employed. Change in NDC areas was assessed against that occurring nationally and within parent local authorities. However, the most useful benchmark is that provided by comparator areas: similarly deprived neighbourhoods in the same local authorities as NDCs, but in non-adjacent wards to avoid potential spillover effects. The same survey design was adopted in the comparator, as in NDC areas. Some 297 people were interviewed in all four waves, who thus collectively represent a ‘comparator areas’ panel. The point should be made here that the comparator areas were not ‘regeneration-free’ controls. They too benefited from support over and above mainstream funding. Yet, on average, NDC areas received more funding than comparators. They were marginally more deprived so they are likely to have received as much non-NDC funding as the comparators; ultimately, few, if any, small localities received the £50 million allocated to each NDC area and it should be stressed that this £50 million was in addition to, not a substitute for, mainstream resources.

Evidence from the household surveys allows change to be addressed in two ways. It is possible to report cross-sectional area-based change (DCLG, 2009a). Data for all 39 NDC, and their comparator, areas can be assessed for 2002, 2004, 2006 and 2008. Yet this evidence is ‘contaminated’ because of population churn from households moving into, within and out of regeneration areas (DCLG, 2009a; Robson et al., 2008). Area-based data thus cannot identify what happens to those who stay in regeneration areas. It seems plausible to assume that more of the change for this group can be ascribed to the effects of the regeneration programme than is the case for area-based data: stayers were open to the effects of NDC interventions for six years: 2002–08. By revisiting previous respondents, it becomes possible to assess outcome change for the NDC panel against those staying in comparator areas.

Secondly, government administrative data were analysed for NDC areas. These covered a range of issues including the location of schools which NDC pupils attended, pupil turnover and educational attainment for NDC pupils when compared with comparator areas, parent local authorities and nationally.

A third strand of evidence emerged from case study work. In 2009, detailed investigations were carried out in four NDCs: Birmingham (Aston), Liverpool, Newcastle and Southwark, selected from areas demonstrating greatest change in educational outcomes. Ten semi-structured interviews were held in each area with education theme leaders, NDC Board members and representatives from agencies, schools and projects. These interviews attempted to unravel factors impacting on outcome change.

NDC Approaches to Improving Educational Outcomes

NDC Partnerships faced considerable problems in relation to educational provision and attainment. Although these varied, they included limited aspirations, parents’ poor experience of school and low attainment. (DCLG, 2010d) In part, these problems were driven by, or associated with, poverty. Across all 39 areas, the percentage of residents whose household was in receipt of free school meals remained unchanged between 2002 and 2008 at 6 per cent: the equivalent national figure falling from 2 to 1 per cent. In addition, rapidly changing ethnic composition raised its own challenges: by 2008, 80 per cent of children in Birmingham Aston spoke English as an Additional Language.

Moreover, pupils attended a large number of schools. Across the Programme, on average, around 80 per cent of NDC pupils attended 10 secondary (post 11) and 10 primary schools (DCLG, 2009b). The remaining 20 per cent were spread across a larger number of schools. Pupils in Hackney NDC in east London attended over 170 schools in 2008. Mobility among children in NDC areas created further complications. Between 30 and 50 per cent of the 2002 primary and secondary school cohorts were no longer living in the same NDC area four years later (DCLG, 2009b). In many areas, instability was accentuated by school closures. In Sheffield, half the NDC’s 2002 primary schools had closed by 2007. Designing interventions to improve educational attainment in neighbourhoods of endemically low achievement was always going to be difficult, but it was compounded by the multiplicity of schools, turnover and closures.

By the end of the financial year 2007/08, Partnerships had spent around £236 million on education, 17 per cent of Programme spend. A further £102 million of match-spend came from other sources. Around two-thirds of this total was allocated to schools, including pupil development and extra-curricular activities (£45 million) and new or improved school and educational facilities (£31 million). Annual expenditure on education increased from £2 million in 2000/01 to peak at £48 million in 2005/06, with revenue accounting for 66 per cent of expenditure. Partnerships adopted various objectives to guide this spend. Southwark formulated its activities to complement the then Labour government’s ‘Every Child Matters’ framework, to ensure that initiatives were consistent with national policy priorities. Newcastle NDC adopted objectives to increase attainment, school attendance and staying on at school rates for 16-year-olds; to enhance parental involvement; to boost participation in lifelong learning; and to ensure the viability of local schools.

Informed by these kinds of objectives, Partnerships funded various interventions, many addressing attainment. Southwark supported ‘Gifted and Talented’, which fostered higher achievement, and ‘The Aylesbury Push’, a study skills service for pupils sitting exams. Partnerships also nurtured relationships with local schools, particularly at primary level, which culminated in various types of support. In one year, Liverpool NDC provided more than 20 support projects to one school including additional specialist teaching and funding to help pupils whose first language was not English. Early years were often seen as a priority. Birmingham Aston supported a SureStart centre offering childcare and early learning experience for young children and job opportunities for local parents. Partnerships also recognised the importance of learning opportunities outside school. Birmingham Aston established a Pupil Guarantee Scheme, based on the principle that all children should have access to curriculum opportunities open to more affluent children such as performing arts and technological learning. Partnerships also promoted adult learning. Birmingham Aston established an Adult Learning Network as a forum for voluntary organisations, colleges and other partners to come together and develop learning opportunities (DCLG, 2010d).

Educational Outcomes: Assessing and Explaining Change

The degree to which these interventions were associated with change is explored next within three themes: change within areas; change relative to other benchmarks; and modelling change.

Change in NDC Areas

Table 1 outlines the educational indicators showing the most change across NDC areas between 2002 and 2008. There were substantial increases in the proportion of residents utilising IT and electronic communications at home and at their place of work or study. Positive change also occurred with regard to educational attainment rates at Key Stages 2 (11-year-olds), 3 (14-year-olds) and 4 (16-year-olds). All of these were significant at the 0.05 level.

OPEN IN VIEWER

Table 1.

Education indicators: the eight showing the greatest change

	2008	Change, 2002–08
Use Internet at home	51	27
Key Stage 4, five or more GCSEs at A* to C	48	22
Use PC at home	57	22
Use e-mail at home	44	22
Key Stage 3 English, level 5	58	12
Key Stage 2 English, level 4	68	11
Use Internet at work/place of study	24	6
Use e-mail at work/place of study	22	5

Notes: Base: all. Change significant at the 0.05 level (Z test).

Sources: Ipsos MORI NDC Household Survey 2002–08; Social Disadvantage Resource Centre.

Benchmarking Change against Other Geographies

As mentioned earlier, assessments of absolute change need benchmarking against what happened elsewhere: nationally, within parent local authorities and in similarly deprived comparator areas. For indicators drawn from the household survey, it is only possible to assess change against what happened in comparator areas. There is then only one example of NDC areas seeing more change over the 6-year period 2002–08 than comparators: a four percentage points relative improvement with regard to those taking part in education and training in the last year (significant at the 0.05 level). So, apparently, the large changes outlined in Table 1 were also occurring in other deprived areas.

The most important indicators here are those reflecting pupil-level educational attainment rates where it is possible to assess change against all three benchmarks. And, as outlined earlier, if one theme dominated educational outcomes it was that of attainment. This evidence is drawn from administrative data and is therefore a complete count of all students at different points in time. Between 2002 and 2008, seven scores in NDC areas improved: English, Maths and Science scores at both Key Stages 2 and 3, and for the percentage of pupils gaining five or more GCSEs at grades A*–C, Key Stage 4. In every case, improvements across NDC areas outstripped improvements at the national level. However, these rates of change need to be treated with caution. Evidence from across the Programme (DCLG, 2010b), indicates that the most deprived individuals and areas saw most improvement: there was more headroom for positive change. It would therefore have been expected that more deprived NDC pupils would have seen more change than occurred nationally. To some extent, that might also have been expected when comparing change against parent local authorities: NDC areas were amongst the most disadvantaged neighbourhoods in each authority. Yet that is only partly true. Of seven indicators of educational attainment, three showed NDC areas improving relative to parent local authorities; for two the gap remained the same; and for two, it widened.

The most important benchmark geography is, however, the comparator areas. Because these are similarly deprived neighbourhoods in the same local authority, they represent the ‘best fit’ benchmark. Improvement was more rapid in the comparator areas for three of these seven indicators, there was no difference for two; and in only two cases did NDCs see more change (Table 2). In all cases, differences were modest—never more than three percentage points over six years. There is no evidence that additional NDC investment was associated with decisive and consistent differences in attainment when compared with other disadvantaged areas in the same local authorities.

OPEN IN VIEWER

Table 2.

Change in educational attainment, 2002–08: NDCs and comparator areas

	2002 (percentages)	2008 (percentages)	Percentage points difference
KS2: English level 4
NDC average	57	68	11
Comparator average	59	73	14
KS2: Maths level 4
NDC average	59	67	8
Comparator average	61	69	8
KS2 Science level 4
NDC average	76	79	3
Comparator average	77	83	6
KS3 English level 5
NDC average	46	58	12
Comparator average	47	59	12
KS3 Maths level 5
NDC average	44	61	17
Comparator average	48	63	15
KS3 Science level 5
NDC average	40	51	11
Comparator average	44	54	10
KS4 GCSE: 5 A*–C
NDC average	26	48	22
Comparator average	28	53	25

Source: Social Disadvantage Research Centre.

In order to assess the validity of these findings, other work was undertaken using a ‘difference-in-difference’ statistical estimation method. In brief, this identified the proportion of pupils in either NDC or comparator areas, attaining a particular educational outcome at 2002 and then again 2007. The treatment groups here were all children living in either NDC or in comparator areas, who sat a particular examination in one of these six years. This amounted to around 26 000 observations in both the NDC and comparator areas. Using multivariate regression models, factors such as prior attainment and school characteristics were taken into account in assessing real differences between these two populations (DCLG, 2010c). This approach therefore determined whether educational attainment outcomes improved for NDC pupils over and above changes that might have been expected in the absence of the Programme. Findings showed that there had been a significant (at the 95 per cent level) positive improvement for NDC pupils in Key Stage 3 science results, but not for other attainment outcomes. NDC pupils did not enjoy better educational attainment rates than peers in other deprived localities. However, there was some evidence that three specific groups of pupils fared better in NDC than in comparator areas: those with low prior attainment at Key Stage 2; those from the lowest income areas; and children from Black Caribbean, other Black and Bangladeshi groups (DCLG, 2010d). For the first two of these, it is likely, as discussed earlier, that positive change reflected, in part, relative rates of disadvantage at the outset: more deprived people and areas made most change. Interestingly, and a theme explored later, there was no relationship between Partnerships spending more on education and positive rates of change in relation to attainment.

Administrative data were also able to address two other considerations which might theoretically have impacted on attainment: geographical spread of students and turnover. However, individual-level outcomes were not significantly better in NDC areas where children could more easily be targeted because of their concentration in fewer schools (DCLG, 2009b). And higher rates of turnover amongst school-aged children were not associated with differences in attainment (DCLG, 2010c). Nevertheless, there is some evidence that high levels of overall residential mobility are associated with poorer educational outcomes. In an analysis of relationships between residential mobility and outcome change in NDC areas, high levels of mobility were found to be associated with lower levels of attainment at Key Stage 4 (DCLG, 2009c).

Discussion

There is little doubt that there was less change for education than for the Programme’s five other outcomes. Apart from ‘undertaken training and education in the past year’, there was little to suggest that indicators moved more positively than in comparator areas. In any event, that indicator is probably best seen as falling within the remit of worklessness, because of the central role of training therein (DCLG, 2009d, 2009e). And when the 39 NDC areas were modelled to help understand why some changed more than others, then, unique to this outcome, spending more on education was associated with less area-level change. These findings are considered later. Before that, two caveats should be mentioned.

First, it needs to be emphasised that across the Programme most change occurred in relation not to people, but to place-based outcomes. It is not possible here to expand on this in detail. In brief, more people will see and ‘benefit’ from place-based interventions such as environmental improvements. Equally, it is easier to encourage residents to become ‘more satisfied’ with their area or accommodation, than, say, to obtain a job or improve educational attainment levels (Lawless, 2011). Even then, however, education saw relatively less positive change than the other people-related outcomes of worklessness and health. Moreover, there is no evidence that ostensibly ‘softer’ education indicators saw positive change either. There was, for instance, a five percentage points increase in NDC residents showing greater ‘trust in their schools’ between 2002 and 2008, but a similar four percentage points rise in comparator areas (DCLG, 2009a).

Secondly, it needs to be recognised that, in common with the Programme’s other people-based outcomes, most of the drivers of educational disadvantage relate to poverty, social class and family dynamics and are not primarily driven by where people live (Cassen and Kingdon, 2007; Dyson et al., 2010; Ferguson et al., 2007; Garner and Raudenbush, 1991; Kerr and West, 2010; Shuttleworth, 1995; Sodha and Margo, 2010; Sullivan, 2001). As others have established, ‘neighbourhood effects’ play only a limited role in explaining, say, teenage educational outcomes (Gibbons et al., 2010) or teenage parenthood (Lupton and Kneale, 2010). To one observer

Neighbourhoods do influence outcomes, regardless of family resources, but … neighbourhoods determine only a small proportion of the variation in individual outcomes … family background matters more’ (Gibbons, 2002, p. 42).

Moreover, “Neighbourhood effects are a non-significant determinant of students’ test score attainments in schools” (Gibbons et al., 2010, p. 36). It was always therefore optimistic to imagine that this Programme of itself would substantially alter area-level educational attainment levels. Nevertheless, it might have been anticipated that more positive findings would emerge. Three themes help to explain the limited nature of change: working with schools; the implications of ensuring spend; and educational investment and change.

First, many Partnerships saw pupil-level attainment as a priority. This inevitably brought into focus relationships between Partnerships and local schools. In all four NDC case study areas, Partnerships established productive relationships with at least some local schools. In Birmingham Aston, Liverpool and Newcastle, NDCs were instrumental in bringing together local head teachers. One NDC officer in Liverpool said that working with schools proved to be a “greater challenge than originally anticipated”, partly because early promises made to schools had been overambitious and it had been necessary to “get the heads round the table again” (DCLG, 2010d, p. 36). It proved consistently easier to work with primary, rather than secondary, schools, because the former tended to be more rooted in their local communities. Newcastle NDC was unable to establish successful relationships with the secondary school and subsequent Academy, with the result that only two projects (Breakfast Club and School Sweatshirts) had been delivered to all secondary pupils. Whilst problems in relation to the former were associated with a previous head, the lack of accountability of the new Academy (either to the NDC or the local authority) had been a disincentive to collaboration. Tensions also flowed from the Programme’s model of governance. In Newcastle, there was evidence of mistrust between community representatives, some of them parents, and practitioners, in relation to the use of NDC resources. Community representatives were loath to fund anything they believed should be the responsibility of the local educational authority and were reluctant to spend money on anything other than school-based projects (DCLG, 2010d).

Underpinning these issues was that tension between attempting to achieve an area-based approach from agencies whose main objective was attaining national targets. Schools were driven by the requirements of the National Curriculum. They were not enthusiastic about having to work in collaboration with other agencies simply because these happened to operate within the same, often arbitrarily defined, ‘place’. Certainly, school heads, used to having final say in resources, were generally resistant to accept what they perceived as another set of targets imposed by NDC Partnerships. As the key local authority contact for Sheffield NDC pointed out

They [school heads] have so little local flexibility … they get driven by their central programmes and all the rest of it … you can understand it would be hard for them, in a tiny bit of the city and a tiny bit of their world (DCLG, 2010e, p. 30).

In their study of Education Action Zones, Halpin et al. (2004, p. 83) similarly concluded that “structural compartmentalisation of agencies” presented an enormous challenge.

Secondly, one issue to emerge surrounds the, apparently perverse, relationship between spend and change: why should spending more on education be associated with negative change in relation to other outcomes? In part, this flows from the nature of the spend process across the Programme. When launched in 1998, it could then be argued that the Programme was an experiment in seeing how 10-year ‘community-driven’ regeneration programmes could transform deprived areas. However, a few years later much of this idealism had disappeared, in part because of the government’s increasing emphasis on delivery through ensuring the spend of annual financial allocations (Lawless, 2006). That placed educational investment, particularly into schools, in an intriguing position. In practice NDC Partnerships often struggled to ensure that financial allocations were spent because projects could be delayed for a range of legal and financial reasons. In that context, spend on schools was often a useful valve. Schools appeared constantly able to assimilate additional resources—for instance, on new teaching assistants. This meant that spend on schools received more NDC investment than would have occurred if the Programme had been able to adopt a more evidence-based approach. This is supported by findings showing differential patterns of spend in those 10 NDC areas associated with most, as opposed to those with least, change. The former, perhaps because they were able to take a more reflective approach, spent relatively less on education than those NDCs seeing least change.

Thirdly, the allocation of resources was skewed too much towards schools. Spend on schools may not be the most effective approach within the intrinsically ‘area-based’ nature of the Programme. At one level, as mentioned earlier, pupils living in NDC areas attended a large number of schools. This raised practical difficulties for area-regeneration agencies wishing to engage with schools: there were a lot of them. The area-based nature of the Programme also worked to accentuate the ‘distancing’ of educational investment from other outcomes. School-based spend was not especially ‘visible’ to most NDC residents, who may anyway not have associated such improvements with their local Partnership. School-focused spend may therefore have limited impact on the attitudes of local people towards the area or the community in which they live. Moreover, in common with the other people-based outcomes of health and worklessness, a lot of what went on directly affected only a minority of residents. It was therefore difficult for household surveys to pick up positive benefits occurring to that minority of local residents gaining from education, or indeed other people-related, interventions. Individuals may well benefit from neighbourhood-level interventions in education, health and worklessness. However, these gains tend not to be picked up at the area level because, in absolute terms, there are relatively few beneficiaries.

Investing in local schools should therefore have been only one element in a portfolio of activities addressing educational inequalities. NDC Partnerships may have been better advised to focus a greater proportion of resources on three alternatives: early interventions; parental support; and non-school-based activity. First, differences in child development and later in educational attainment begin to emerge even before entry into school (Carter-Walland Whitfield, 2012; Duncan, Smith and Allen, 2008).

Later interventions to help poorly performing children can be effective but, the most effective and cost effective way to help and support young families is in the earliest years of a child’s life’. (Field, 2010, p. 5).

There is value therefore in identifying young children warranting additional support. Secondly, evidence suggests that parental support for children’s learning has a major impact on attainment (DfES, 2007a; Harris and Goodall, 2007; Goodman and Gregg, 2010), particularly when it takes place in the home. Studies suggest that family background plays an influential role in educational attainment (Ferguson et al., 2007; Garner and Raudenbush, 1991; Shuttleworth, 1995; Sullivan, 2001). In effect

Children from lower socioeconomic groups may have different background knowledge, skills and interests which are not reflected in the school curriculum; and are less likely to have the kinds of social connections which offer inspiration and opportunities’ (DCSF, 2009b, p. 57).

There was always therefore a case for NDCs to allocate a greater proportion of educational investment in supporting parents and children in the home. And, thirdly, children’s experiences outside the school system also affect educational outcomes (Wikeley et al., 2007). Supporting projects in these three areas would also have reaped benefits for Partnerships. It is reasonable to argue that allocating a larger proportion of educational investment to non-school interventions would have been more likely to help generate more positive outcomes. And Partnerships would have had more control over the nature of interventions.

However, and to bring the debate round full circle to the ‘community-driven’ nature of this ABI, adopting a more evidence-driven, ‘extra-school’ education strategy was always unlikely to be adopted within a Programme wherein local residents constituted a majority on Partnership boards. For local residents, there were ‘natural’ allies with whom it made sense to collaborate for each of the Programme’s outcomes: the police for crime, local health providers for health and, in relation to education, local schools. These alliances had some unfortunate consequences. Partnership boards too readily accepted that these agencies were the most appropriate organisations with which to work, a bias which underestimated the degree to which the latter were generally wedded to achieving national, not local, objectives. NDC boards also too easily assumed that such agencies were aware of the key evidence base in relation to ‘area-level’ interventions. And focusing on these few key delivery agencies tended to marginalise thinking which located outcome change within wider contexts

Area-based approaches to educational regeneration cannot hope to succeed if they ignore the macro-political context in which they operate (Halpin et al., 2004, p. 84).

Concluding Comment

Findings developed in this paper point to the problematic narrative surrounding education in the NDC Programme. Change was relatively limited: other deprived areas generally performed at least as well; and there is evidence of an inverse relationship between educational spend and change. Several reasons can be used to help explain this pattern. However, the most potent explanatory factor probably surrounds the inclination of NDCs to prioritise support for schools in order to address questions of attainment. It is entirely understandable that parents living in disadvantaged neighbourhoods, served by apparently poorly performing schools, should aspire to something better for their children. Interventions were strongly influenced by the perceptions of local community representatives who saw local schools as having the major role in improving educational standards. In effect, NDCs spent too much, of an already relatively ‘inflated’ educational budget, on local schools. There always was a strong argument that investing in a wider range of non-school interventions would have helped to target resources more effectively on children and households in greatest need. This throws into sharp relief a dilemma central to area regeneration and one which is likely to impact on notions designed to create a ‘Big Society’ (Kisby, 2010). It seems a sensible and progressive step to provide local citizens with more say on what goes on in their area, an objective central to the NDC Programme. Yet what happens when popular sentiment, here articulated through community representatives on NDC Boards, sits ill at ease with the evidence base? Had NDC investment been more tightly aligned with evidence as to what might work, there would have been relatively less emphasis on providing schools with ‘support’ and relatively more directed at supporting younger children and their parents, combined with targeted ‘out of school’ programmes of support. Had such an approach been adopted, it seems plausible to argue that outcomes may have proved more positive. Ultimately, however, as explored earlier, because change is largely driven by non-spatial factors such as socio-demographics and levels of deprivation, ABIs are unlikely to preside over substantial change in relation to any people-based outcome.