Identifying deprived neighbourhoods is an obvious basis for multiple-level interventions for community development, addressing both structural and individual problems, in order to enhance children’s physical and mental health, and cognitive development. This was the basis for a programme initiated by the British government, called National Strategy for Neighbourhood Renewal (Glass, 1999; ODPM, 2005; Eisenstadt, 2011) which aimed over 20 years to regenerate all of Britain’s highly deprived local neigh- bourhoods (which constitute about ten per cent of all urban neighbourhoods identified at the “voter enumeration district” level). Sure Start programmes, as part of this initiative, were established on an area basis (Eisenstadt, 2011; Belsky & Melhuish, 2007) using census data to identify urban areas with populations potentially at risk.
The neighbourhood regeneration programmes were greatly diminished after 2010, when the New Labour government was replaced by a Coalition, and then by a Conservative government. Sure Start itself has taken a little longer to be discarded by central government, since phasing out the many neighbourhood child care centres was politically difficult, so the reduction and changes in funding affecting these centres has been more gradual (Sammons et al., 2015 a & b) .
The aim of this policy review is to examine the numerous challenges to the Sure Start initiative-methodological, political, and organizational―which surrounded a major public policy initiative on behalf of children up to age five. As Melhuish et al. (2010) acknowledge in their overview of the first phase of Sure Start a change of national government in Britain meant that ideological pressures meant that the focus, design and reporting of Sure Start had to change within a changing political climate, and identification of “success” in programme delivery may not have been welcome news for subsequent UK governments for whom reducing public expenditure, rather than enhancing the life chances of children born into the poorest social classes, even though this initially expensive venture would have been highly cost effective in the medium and long- term (Allen, 2011) .
2. Initial Evaluation Studies of Sure Start
The New Labour government of Britain had paid some attention to the abundant medical and social evidence on the corrupting, demoralizing and demeaning effects of chronic poverty on family life, and on children’s health and welfare (Bagley & Sawyerr, 2008) . The New Labour government thus initiated the Sure Start programme in 1998, as part of its goal of halving the incidence of child poverty by 2010.
The declared goals of Sure Start were: “To work with parents-to-be, parents and children, to promote the physical, intellectual and social development of babies and young children―particularly those who are disadvantaged―so that they can flourish at home when they get to school, and thereby break the cycle of disadvantage for the current generation of young children.” (Sure Start Research Team, 2008) . This initial three-billion pound programme, modelled to some extent on the American Head Start initiatives, aimed to provide improved parenting skills in areas of high deprivation, focussing on the first five years of a child’s life (Barnes et al., 2005) . Unfortunately, the systematic integration of Sure Start with various medical interventions was dropped following the initial pilot work, largely on grounds of cost, although such integration did remain (and was shown to be highly effective) in some centres (Sammons et al., 2015a) .
The initial workings of Sure Start (in the integrated model, using medical, social work and educational resources) were described, for example, in an evaluative study in the North West region of England (Pearson, 2005) . Within the selected areas, participant families were identified and referred by community midwives, and the programme offered support to parents (particularly mothers) to improve their health and emotional and social development, and their parenting abilities. In addition to group sessions for effective parenting before and after the child’s birth, parents were usually offered a maximum of four individual counselling sessions, although further sessions might be offered for families considered at high risk of neglect or abuse of children. Involvement in the programme was voluntary, and in the settings studied by Pearson (2005) some 70 per cent of parents approached initially agreed to participate. However, of those parents considered most at risk for “problem parenting”, 50 per cent chose not to attend any of the individual counselling sessions, and less than a quarter completed all four sessions. Among the reasons given for not attending were “illness of self or family member”.
Fathers were particularly difficult to engage, and because evening sessions were not usually offered, parents working full-time often had difficulty in attending. The initial evaluation of this programme was qualitative rather than quantitative, and there were few indicators of outcome, apart from the fact that most parents who had participated said that the experience had been enjoyable and positive. But this kind of “halo effect” is common in evaluation work, and merely tells us that those who participated fully in a voluntary programme were probably those least likely to have required such a service.
Attached to the national Sure Start programme was a major evaluation programme based at Birkbeck College, University of London. This team first of all, examined service delivery to 15,000 families and their focus child in 150 Sure Start nursery centres in order to provide a description of services actually delivered. Secondly the team attempted to assess whether children, families and communities had actually benefited according to various indicators. Twenty six centres were randomly selected from the 150 centres for intensive study, children and families in these centres being compared over six years with initially similar families in fifty “Sure-Start-to-be” comparison areas (Eisenstadt, 2011) .
Belsky et al. (2006) published details of the first statistical evaluation of Sure Start, based on interviews and tests involving 3927 mothers and their children who were enrolled in the programme, at the age nine months and three years. The target group were compared with 1509 mothers and children from similarly deprived neighbourhoods, who were not yet enrolled in Sure Start. The main dependent variables were mother’s perception and use of community services; her family functioning; her reports on her child’s health and development; and a measure of the child’s verbal skills at age three.
The results of this initial evaluation were disappointing: differences between target and comparison groups were small, and when statistically significant pointed to adverse outcomes for the most deprived mothers and children enrolled in Sure Start. Children of teenaged, single mothers, and unemployed single parents who participated in Sure Start had children with poorer verbal ability in the third year of life. Sure Start had the most beneficial effects for the least deprived, intact families living in areas with lower levels of deprivation. Apparently these mothers were able to elicit additional helping and support networks unavailable to the most deprived mothers. Overall, outcomes were slightly better in Sure Start programmes which were delivered within a health services framework. A follow-up of a pre-2003 Sure Start cohort into the early years of schooling showed that the focus children had better social skills, but were no better at scholastic attainments than were control children (Schneider, Ramsay, & Lowerson, 2006) .
The Minister for Children and Families defended the Sure Start programme, arguing that positive outcomes should be seen in the longer-term, rather than in the first few years of the programme (Hughes, 2005; Redfern, 2005) . The authors and evaluators of Sure Start might have been looking to the evaluations of the US Head Start programme, which also showed few short-term benefits, but nevertheless showed highly significant gains for the child participants when they were in their teens―in terms of school achievements, adaptive behaviours, and educational and occupational aspirations and achievements―compared with controls (Oden, Schweingart, & Weikart, 2000) .
The need for a fully effective programme which could fulfil the idealist goals of Sure Start was underlined by the longitudinal research by Joshi and colleagues (Joshi, 2007) . This study used data from the Millennium Cohort of 15,500 British children born in the years 2000 to 2002, and indicated that children from the most socially advantaged social groups were on average, a year ahead of children from the least advantaged group on the School Readiness Test, which assessed a child’s recognition of words, numbers, shapes and colours, regardless of whether or not they had been enrolled in preschool nurseries. This series of studies was unable to show that Sure Start programmes had been effective in enhancing “school readiness” in children from the most disadvantaged families. A further report from the Millennium Cohort in 2008 showed that before they entered schools, children of young, poorly educated mothers were nearly a year behind in the their vocabulary scores, a difference that increased for each year that they remained in school (Joshi, 2008) . In this study, boys with conduct behaviour disorders, with depressed and often punitive mothers, were the most disadvantaged in terms of reading readiness, and it was clear that these were mothers and children whom Sure Start should focus on in particular.
It may be countered that Sure Start focused not on cognitive goals, but on parenting capacity and the development of behavioural and emotional competence in children. Ideally of course, cognitive and emotional goals should be simultaneously addressed in a comprehensive programme for the most disadvantaged families (Eisenstadt, 2011) .
One problem which emerged in evaluative studies of Sure Start was that of integrating the work of health care, social work, child care and clinical psychology specialists involved (Anning & Ball, 2008; Avis et al., 2007, Edgley & Avis, 2006, 2007) . Apparently programming in some areas was working better than in others, and this could have been due to varying degrees of integration of the professionals involved, or of the differing nature of the communities in which intervention was attempted (Barnes, 2007; Melhuish et al., 2007; Melhuish, Belsky, & Barnes, 2010; Raymond, 2009) . In addition, some severely disadvantaged clients may have felt stigmatized by the proposed interventions, accounting for their low take up of services (Coe et al., 2008) . Failure of Sure Start programmes to recognize, or intervene with severe maternal depression (especially likely for single, abused or deserted mothers) was another problem which could be associated with reduced impact (Raymond, 2009; Coe et al., 2008) . Another identified problem was that some Sure Start centres were failing to link effectively with black and other ethnic minorities (Craig, 2007) .
The planned expansion of Sure Start centres after 2004, from some 1400 to 3500 over ten years (Eisenstadt, 2011) , faced the problem that not enough qualified workers were readily available to staff such expansions; and the budgetary allocation for Sure Start appeared to be inadequate for training such new staff (NAO, 2007; Anning & Ball, 2008) .
A useful policy analysis by Gray & Francis (2007) goes some way to explain both positive and negative aspects of Sure Start’s initial roll-out phase. They draw specific lessons from a comparison of Sure Start with the American Head Start programmes. Their analysis offers the following conclusions:
1) Early interventions, as the American experience shows, can significantly improve the life chances of many children throughout their lifespan; but failure to provide adequately expanded funding can impair both the quality and impact of early intervention programmes of this type. These positive outcomes occurred despite some differences in the American and British models.
2) There is a temptation for evaluators to focus on narrow, measurable objectives; but this runs the risk of ignoring broader aspects of success, and a combination of quantitative and qualitative evaluation techniques may be needed.
3) Programmes must be flexible in meeting local conditions, and the needs of individual families, while remaining faithful to the original programme goals.
4) Be aware that multiple programme objectives may conflict with one another, and political demands to divert early intervention programmes to meet new or multiple goals should be avoided.
5) Evaluation may show that the programme works better with some client groups, and in some areas. The failure to be effective with all client groups should not be seen as a general failure of the programme in it its initial years.
6) The English and Welsh Sure Start programme was probably rolled out too fast, in order to fulfil political goals. Funding, although initially generous, failed to recognize problems of recruiting and training staff for a programme which was, at that stage, unproven.
7) Now that Sure Start was entering its second phase and building on experience, it was crucial that funding matched the needs of what was still a developing programme. Failure to fully fund the programmes because of for example, a recession and cutbacks in public funding, could be disastrous for the long-term success of Sure Start.
The methodological challenges facing the Sure Start evaluation team should be acknowledged, and Melhuish et al. (2010) in their final report on the first phase of Sure Start clearly acknowledge these challenges. The methodological design, influenced by the best standards of educational psychology, using validated measures analysed by the latest statistical techniques is laudable in theory, but may have problems with large- scale data sets, since the researchers have to rely on non-research trained field workers to collect data. Thus the reliability of data may at times be questionable; and certainly, minimally trained field researchers were not “blind” to the research setting (Sure Start versus controls) when they administered instruments and collected data. Melhuish and his colleagues (2010) coped with this problem to some extent by subsampling groups of Sure Start centres for more intensive and controlled investigation.
It is of importance that the distinguished scholar, Michael Rutter (2007) who had conducted many large scale epidemiological studies in child psychiatry, praised the Sure Start team led by Melhuish and Belsky for their careful methodology, and accepted the overall validity of their findings.
3. Sure Start’s Developing Success
Notwithstanding the earlier problems of programme organization and service delivery, Sure Start seemed to have “bedded down”, gaining a “second wind”, as evidenced by later evaluation studies (Edwards et al., 2007; Melhuish et al., 2008, 2010) . Overall evaluation of the Sure Start programme when the children were aged five-plus, according to Melhuish & the Sure Start Research Team (2010) provided rather more optimistic findings than the 2005 evaluations. There were now more than 9,000 families invo- lved in SSLPs (Sure Start Local Programmes) in 150 areas. Comparison between SSLP participants, and matched non-SSLP families and children enabled a wide range of family and area background factors to be controlled. The main findings were:
1) Parents of 3-year-old children in the programmes showed less negative parenting, while providing their children with a better home learning environment.
2) Children in SSLP areas had better social development, with higher levels of positive social behaviour and independence/self-regulation.
3) The SSLP effects for positive social behaviour appeared to be a consequence of en- hanced parenting behaviours.
4) SSLP children had higher immunization rates and fewer accidental injuries.
5) SSLP families used more child and family-related services.
6) Positive effects associated with SSLPs applied to all of the participants, rather than to different subgroups identified in 2005.
7) The more consistent benefits associated with SSLPs in 2008 compared with 2005 might well reflect the greater exposure of children and families to the programme, and to the evolution of a more focussed and sophisticated type of programme delivery.
Sure Start, like the American Head Start programme (Currie & Thomas, 1993; Bennett & Hustedd, 2005; Cameiro & Ginja, 2014) might have global advantages which spread out from earlier gains, reflected in better achievement in later years. The American programme found that by their mid- to late-teens the children enrolled as infants made better school progress, dropped out of school less, were more likely to go on to college, were less delinquent, and less often became pregnant. These gains made the early investment in Head Start highly cost effective (Cameiro & Ginja, 2014) .
Further evidence that Sure Start Local Programmes (SSLP) were learning valuable experience over time, came from the longer term evaluation programme by Melhuish and his team (Melhuish et al., 2008, 2010; Melhuish, Belsky, & Barnes, 2010) . In a quasi-experimental study which compared 5883 3-year-olds and their mothers who were enrolled in Sure Start nurseries, with a comparison group of 1879 3-year-olds of similar backgrounds, not enrolled in Sure Start. The Sure Start children had statistically significant advantages in the following areas, after all relevant background factors (e.g. family size, presence of father, dependence on financial benefits) were controlled for: better social behaviours; more self-confident independence in the child; less negative parenting; better home learning environment; more use of relevant family support services. These advantages held across different regions, ethnic groups, and social class backgrounds.
However, the SSLP children had no significant advantages in several other desired outcomes: mothers smoked as much as before; children’s language skills were similar; mean BMI indicators (predictors of obesity) for child and mother were similar across the two groups; father’s involvement was no greater; personal life satisfaction was no greater; and mothers in both SSLP and control populations, often rated their housing and urban environment negatively. A similar proportion of children from both groups still had incipient behaviour problems. It remained to be seen whether prolonged exposure to SSLPs, and added programme experience and feedback based on evaluations such as these could yield better results in the longer term.
The medical focus of Sure Start had been emphasized in a successful intervention with parents in deprived areas whose children were at risk of developing conduct disorder (Hutchings et al., 2007) . In this controlled study 153 parents were offered behavioural support and focussed counselling to help them cope with their child’s incipient problem behaviour. Results showed clear and significantly different positive outcomes for children in the focus families, compared with those in the wait-list controls, in terms of reduction of problem behaviours.
This important new direction for Sure Start was emphasized by further work of the team led by Hutchings, Bywater & Daly , and Bywater et al., 2009 , and came from a follow-up of this experimental programme, based in Wales and North West England which identified children at particular risk of developing conduct disorder (and later delinquency) because of their identified symptoms of Attention Deficiency and Conduct Disorder (ADHD) at age three. This team identified 50 children with serious levels of ADHD and instructed and monitored their parent(s) in giving appropriate feedback to the child in ways which reduced the chronicity of symptoms. The approach is similar to that described by Bagley & Mallick (2000) of providing “goodness of fit” between child behaviour and parental feedback in ways which lead to the “spiralling down” of difficult behaviour to normal levels. Jones et al. (2008) achieved an improvement of 57 percent in the focus group (criterion, falling below the clinical level as indicated by scores on the Connors Rating Scale) compared with 21 percent in the untreated, waiting list controls. These gains were maintained, in comparison with controls, at follow- ups 12 and 18 months later.
These interventions were shown to be clearly cost-effective (Bywater et al., 2009; Jones et al., 2008) . Scott (2007) commented, on the basis of this and earlier research, including Scott et al. (2001) , that although these interventions were relatively expensive (about £1,800 per family), in the long run these interventions could be very cost effective, given the known costs of children who enter cycles of juvenile delinquency and rebellion in school and community. Sure Start could be most effective not principally as a service agency, but also as a screening agency which refers for intensive help families with children most at risk.
Melhuish, Belsky & Barnes (2010) summarised their evaluations of the first phases of Sure Start in the following terms: Sure Start has been evolving, and ongoing research has partly influenced this process. Later developments have considerably clarified guidelines and service delivery. It is plausible that the improved results in the evaluation of Sure Start reflect actual changes in the impact of Sure Start programmes resulting from the increasing quality of services, greater attention to the hard to reach, the move to children’s centres, as well as the greater exposure to programmes, of children and families in the latest phase of the impact evaluation. The results are modest but suggest that the value of Sure Start programmes is improving. The identification of factors associated with the more effective programmes has propelled recent improvements in Sure Start Children’s Centres and may be in part the reason for the improved outcomes for children and families now found in Sure Start. (p. 160)
4. The Second Stage of Evaluation: The Oxford’s Team Results
Eisenstadt (2011) , the civil servant responsible for overseeing the setting up of Sure Start in the early years of the New Labour government, gives an insight into the fierce competition in bids to evaluate the early years of Sure Start-the awarding of the research contract was certainly valuable for the university or research group which took on the task, and many influential publications were likely to follow. Indeed, there was “controversy” at the contract being awarded to Birkbeck College, University of London, perhaps because these scholars promised a quantitative form of evaluation, rather than a more global, qualitative perspective (although the two perspectives are of course complement rather than compete with one another). The social psychiatrist Michael Rutter (2007) was asked by government to conduct an independent audit of the earlier research emerging from the Birkbeck group, and he observed: “Given the constraints imposed by government, this was a rigorous and careful an evaluation as could be undertaken.... The research team are to be congratulated on their high quality research. As a consequence, there is every reason to trust the research findings.” (pp. 197-209).
The second major contract for the evaluation of Sure Start, now relabelled for political reasons as Children’s Centres (Lewis, 2011) was awarded to a group at the University of Oxford. In the event, this group carried out a methodologically careful quantitative approach, reflecting the rigours of research in educational and clinical psychology. The team was joined by Melhuish, who had been a key member of the Birkbeck group. And Eisenstadt, having retired from her civil service post, joined the Oxford group, writing an extremely interesting account of the political sociology of Sure Start in her (2011) book: Providing a Sure Start: How Government Discovered Early Childhood.
The major series of reports on the second wave of evaluation were issued in 20151 under the authorship of the four key researchers, with acknowledgments to a number of specialist or consultant researchers (Sammons et al., 2015 a & b) . These new evaluations did not use the term “Sure Start Centres”, but instead termed their work on centres by the new, politically correct terminology as “ECCE”, meaning “Evaluation of Children’s Centres in England”. The research was divided into five “Strands” for the period of evaluation, 2009 to 2017, and a number of evaluations were to be published over this timespan.
The Strands were:
Strand 1: Delivery and use of ECCE in 509 “most disadvantaged” areas, 2011 to 2013.
Strand 2: Interviews with a sample of 2,608staff in 128 ECCE centres in 2013.
Strand 3: Reports of visits to a sample of 121 of the 128 “focus” ECCE centres (Goff et al., 2013) .
Strand 4: The impact of ECCE programmes on child and family functioning in 1,305 boys and 1305 girls attending 117 of the 128 focus centres, studied at three points in time, when children were aged between 9 months and 38 months2.
Strand 5: Cost-benefit analyses (Briggs, Kurtz, & Pauli, 2012) ; and a yet unpublished, later report.
The Strands 1 and 2 evaluations focussed on a broad range of non-child outcomes examining the underlying goals of Children’s Centres “…to support all children and families living in particularly disadvantaged areas, by providing a wide range of services tailored to local conditions and needs.” (Sammons et al., 2015a)
Results from the Strand 1 evaluation reported: “…high levels of parent satisfaction, and clear evidence of improved ‘personal, social and emotional development’, with 92 percent of mothers interviewed being ‘very happy’ with the programmes offered.” (Evangelou et al., 2014) . The delivery of services on a neighbourhood basis was offered to a broad range of families, including those not currently experiencing material or social stress. The aim was to avoid stigmatising low-functioning families, through services which were inclusive of the whole community (Sammons et al., 2015 a & b) . Nevertheless, within this broad range of services, those “most in need” were identified, so that for such families a more focussed approach could be offered, in terms of mothers’ material and mental health problems, problematic parent-child relationships, and material difficulties (Lord et al., 2011; Sylva et al., 2015) .
The major evaluation, Strand 4, was based on longitudinal data collected at three points in time in the pre-schoolers’ lives (from 18 months through to four years of age), using a number of validated measures of children’s behaviour and cognition, parent-child interactions, and maternal and family functioning. Among the measures was one intriguingly termed the CHAOS scale which measured family “parental distress and dysfunctional parent-child interaction”, based on a validated measure using in American Head Start projects (Matheny, 1995) , estimating (through mother’s self-re- port) ‘confusion, hubbub and (dis) order’ in family life.” Overall, 13 child, mother and family outcomes were measured. The longitudinal design allowed the researchers to identify what services were available and were used, and their possible outcome (and influence) over a 30-month period.
The broad research question was: “What aspects of children’s centres (management, working practices, services offered, services used) promoted better family, parent and child outcomes?” (Sammons et al., 2015a)
The researchers used advanced statistical modelling (based on multiple regression analyses) which allowed them to partial out the influence of any predictor variable on any outcome variable, all other factors controlled for. The demographic and health factors controlled for included the family’s socioeconomic status, ethnicity, family size and birth order, maternal age and education, child’s perinatal health status, degree of neigh- bourhood deprivation, and mother’s initial mental and physical health status.
Significant predictors of negative CHAOS scores (Matheny, 1995) and related parental distress indicators when the child was aged three were (all other factors controlled for): Mother’s poor mental and physical health when first interviewed; family’s lack of material resources or chronic unemployment at the outset; larger families; and mother’s low educational achievement. This model applied to mothers of both genders, but was more marked in the mothers of preschool boys.
The ECCEs did have a significant influence on problematic family functioning in the focus child’s fourth year of life. The more that the families used the child care centres, the less likely were they to have negative outcomes in terms of CHAOS scores, and problematic parent-child interactions. When a health visitor was based in the ECCE centre, outcomes were also better. The better the worker-child staff ratio in a centre, the better the outcome, but only when the centre was operated on an “Educational Leadership” model, with kindergarten classes run by qualified teachers (Sammons et al., 2015b) .
The period over which the research was undertaken was one in which Sure Start funding was being offered on a changing legal and fiscal basis, and this resulted in both some contraction of services offered in CC centres, and the closure of a number of centres: Sammons and her colleagues (Sammons, 2015a) observed: “Since 2010 [Sure Start] child care centres have experienced considerable turbulence and volatility as a result of changing organisational models, funding constraints linked to budget cuts, and addressing new children’s centres ‘core purpose’. Local authorities were given responsibility for making decisions on which services were most required per locality. The ring-fence for Sure Start funding was removed and the Early Intervention Grant (EIG) was introduced in 2011, so it is not possible to put a figure on central government funding for Sure Start from 2011-12 onwards… and many children’s centres reduced their services.” (p. 2).
5. Government Policies Undermining the Funding Base, and the Effectiveness, of Centres
A survey in 2010 by The Day Care Trust, an independent charity (Bennett, 2011) of 3,578 Sure Start children’s centres in England, found that more than 2000 had reduced the programmes offered because of diminished funding in the previous year: some 250 centres had already closed, or were planning to close in the forthcoming financial year. These reductions in service reflected two factors: transfer of financial responsibility for the Centres to local authorities from national government; and a cumulative reduction in local authority budgets of 7 percent per annum.
By 2015, in England, Sammons and her colleagues (2015b) reported, that another 142 children’s centres had closed, leaving 2816 centres remaining in the country. Between 2010 and 2013 government funding for the Sure Start initiative had fallen by 28 percent; it was not possible to calculate the extent of funding cuts in England after 2013, because the changes in the “block grants” to local authorities did not allow this fiscal breakdown, although spending in the national health service budget on child and adolescent mental health also reduced by around five percent over a 4-year period (Sammons et al, 2015b) .
The evaluation research by the Oxford University team found that when child care and other services were reduced in quantity and quality during the period of their 3- year evaluation, outcomes in terms of parental distress, and poor parental functioning were significantly increased, compared with centres which maintained the full range of staffing and services. The best outcomes for parents and children were delivered by centres which offered not only a stable and fully staffed service, but which also linked to health care supports, and were led by an educationally-trained professional.
Not surprisingly, parental distress and high-CHAOS families tended, overall, to have children with poorer cognitive and emotional outcomes-effects which were stronger for boys than for girls. However, overall cognitive gains or deficits resulting from family and children’s centre variables were marginal. The strongest effects were those concerning externalising (e.g. conduct) behavioural disorders, and internalising (e.g. anxiety) problems. The “neediest families” in contact with Centres clearly got the most help. For most (but not all) families this paid dividends in terms of positive outcomes, particularly for boys at the threshold of serious conduct disorders (Sammons et al., 2015a) .
Health visitors (combining the role of public health nurse and social worker) were particularly effective in helping mothers who had problems with alcohol or drug abuse. Being in financial distress was systemic in causing various family dysfunctions, and when intervention was able to stabilise a family’s finances, maternal mental health and mother-child relationships also improved, as did child behavioural outcomes. The greater the level of family disadvantage, the more likely was the family to use the services offered by the child care centres. It is implied that if the Centres were more generously funded, and could offer a wider range of specialist services, then outcomes for children would have been more favourable. Centres which employed a multiagency model, linking with health, social and educational services had the best outcomes.
There was one negative finding however. Mothers whose social and mental health and material conditions, and lack of partner support were profoundly negative at their first contact with a Centre, actually deteriorated over time (to the disadvantage of their children) despite their high level of dependency on anything the Centres had to offer. This did not mean that the Centres were making things worse for some families. Rather, it implied that a small fraction (perhaps five percent) of mothers needed more intensive and more highly skilled help than could be offered by even the best-staffed Centre: “…Children’s Centres typically did not have highly qualified specialist staff to support complex mental health or social problems.” (Sammons et al., 2015b: p. 23) Furthermore, according to the evaluation team “Cuts to mental health services further hit child care centres’ ability to treat and refer those with complex social or mental health problems.” The five percent of mothers whom Sure Start did not help had very poor physical and mental health at the outset, had rarely succeeded in school, were not usually supported by a stable partner, and endured chronic material poverty. Their mental health and their parent-child relationships steadily deteriorated over a two-year period, this decline being particularly marked when the children’s centre services were reduced or withdrawn.
The research found that 14 of the 117 centres studied had experienced cuts in their budget, or loss of staff. However, because of local authority support 32 of the centres were actually expanding services, and child and parent outcomes were most favourable in these expanding centres, and least favourable in the contracting or underfunded cen- tres. Moreover, the negative effects of reduction or withdrawal of services for highly- stressed families was significantly greater than the overall positive effects of continued services. It appeared that programme reduction or withdrawal acted as an additional stressor in the lives of already highly-stressed families (Hall et al., 2016) . This has been emphasized in a qualitative study of Sure Start child care centres in Liverpool (Campbell et al., 2016) which showed that as child poverty was increasing in areas of deprivation, available services including Sure Start centres, were experiencing (as were other social and health services for children and families) a reduction in funding, making child and parent outcomes appreciably worse.
The main evaluation study by the Oxford team (Sammons et al., 2015b) acknowledged that although moderately disadvantaged families who made maximal use of the facilities offered, did seem to be “on track” for improved life chances, in general the Sure Start programme had not altered the overall impacts of inequality on child and family health in the most disadvantaged sectors of population in England. Very disrupted and disturbed families were often not helped by the children’s centres, and in an era of reductions in social work and mental health programming, a likely outcome was that many of these children would eventually be removed into care, since resources which could support these families were not available. Families who had additional resources of “social capital” seemed to have gained most from Sure Start (Bagley, 2011) .
Sure Start child care centres were established in areas serving “mixed” communities, both in terms of economic and ethnic status. The Oxford team’s evaluation sample reported that 71.1 percent of families served were traditional “white British”, and these included most of the “very deprived” families. Some 12 percent of the children served came from black or “mixed-race” backgrounds. Mothers and children from these backgrounds were not identified as having particularly poor or good outcomes in terms of social adjustment, cognition and behaviour. This underlines the fact that the neigh- bourhood-delivery model of Sure Start would inevitably offer services to some families who were climbing out of poverty, or indeed did not require child care other than for enabling mothers to pursue career options. Although Sammons and colleagues (2015b) did not specify ethnic profiles of the economically poor and disrupted low-functioning families, we infer that these are largely inter-generationally disadvantaged families of “poor whites”, in whom problems of children are likely to influence the development of those children when they are adolescents and young adults, and have children of their own.
Sammons et al. (2015b) compare their evaluations with those of the earlier Sure Start (NESS) evaluations, and concluded: “The ECCE results support and extend those of the earlier NESS study. They demonstrate that children’s centres do have the potential to promote better outcomes for families and to a lesser extent, for children who are engaged in specific programmes (such as high quality childcare). At present the focus is on family and parenting services, and perhaps unsurprisingly, such outcomes show more evidence of impact in this evaluation.”
But funding and service changes since 2010 were clearly making evaluation using straightforward statistical models extremely difficult. It could be however that in the very long-term, follow up of Sure Start and Children’s Centre “graduates” will show that as young adults, they will manifest some positive gains in educational achievement, employment, and stable family formation-as demonstrated in the American Head Start research (Currie & Thomas, 1993) .
Offering a comprehensive programme of child and family support (including income and housing support) in neighbourhood family centres, with links to local schools, and offering individualised support for each child is an ideal―and the Oxford team point, enviously, to the “universal care” models in Finland (and also in Norway) (Kekkonen, 2015) . Such programmes are generously funded, and comprehensive in nature and are linked to Finland’s overall programmes of income equality, and the high scholastic achievement levels of children and adolescents later on (Bagley & Sawyerr, 2008) .
The English Sure Start, in comparison with this Finnish model, is poorly funded and subject to the political forces which prevail in a country in which neo-liberal economic and policy models dominate social care options. The emerging conservative-liberal alliance in 2010, and the subsequent conservative government rebranded Sure Start as Child Care Centre programmes, and withdrew the evaluation contract from the University of London.
Sure Start child care centres began, idealistically, as neighbourhood programmes but also offered services to families and children who did not actually need those services. The evaluation programme focussed on families using the services, rather than on families who were in greatest need, but who might have not accessed the services offered. The kind of school-based research identifying particular families in whom problems are intergenerational in nature, which we applied in a government-funded demonstration project (Bagley & Pritchard, 2008 a & b) would, in contrast, result in comprehensive services being focussed on those most in need. There is irony in that the UK government appears to have ignored the obvious benefit of investing in disadvantaged families by focussing on schools in disadvantaged areas, identifying children and families who were clearly in need of supportive intervention using a much cheaper option than that employed in Sure Start, which offered services to families who were clearly not in need of social and psychological supports. This alternative model (focussing on particular schools, and on sub-groups of “problematic” children and their families, including the provision of services for pre-schoolers) (Bagley & Pritchard, 1998 a & b; Williams & Pritchard, 2006) has been shown to be highly cost-effective. The follow-up studies of “graduates” of these school-based programmes have shown that by late adolescence, these young people have significantly fewer indications of problematic behaviour (e.g. criminality, substance use, unwed pregnancy). In the medium term these programmes are highly cost-effective (Williams & Pritchard, 2006; Bagley & Sawyerr, 2008) .
The Sure Start/Children’s Centres programme was, paradoxically too ambitious and idealistic, and spread its limited funding over too large a client group. An alternative model would have been to use the primary school and the neighbourhood it serves as the centre of social work intervention, identifying pupils with problems of learning and behaviour, and then offering intensive engagement with their families and their other children, with special support from teachers and therapists (Bagley & Sawyerr, 2008) . Ultimately, as Standing (2014) advocates, the problems of the very poor can only be successfully addressed in the longer term within the context of a stable and generous “basic living allowance” available to all citizens. Evidence continues to emerge that providing low income families with a stable income can have significant public health benefits (Gray, 2013; Muenning et al., 2016) . And as the Allen Report urged the UK government in 2011, influences on a child in the first five years of life are very hard to undo, and can have negative impacts throughout adult life, at great cost for a variety of service providers. The same model also shows that enhancement of the material, social and psychological circumstances of a child’s development in these crucial preschool years, can be highly cost-effective (Melhuish, 2016) .
1The Reports were released by the Department for Education on “the night before Christmas” in December 2015, allegedly so that they would attract little press attention (Guardian, January 14, 2016). In the event, the press did, by and large, ignore the Reports.
2Fathers were not interviewed, since they are difficult to interview when they are working full time; or when they are absent from the family.
 Avis, M., Buchanan, D., & Leighton, P. (2007). Factors Affecting Participation in Sure Start Programmes: A Qualitative Investigation of Parents’ Views. Health, Social Care and Community, 15, 203-211.
 Bagley, C. (2011). From Sure Start to Children’s Centres: Capturing the Erosion of Social Capital. British Journal of Educational Psychology, 37, 95-113.
 Bagley, C., & Mallick, K. (2000). Spiralling up and Spiralling down: Implications of a Long-Term Study of Temperament and Conduct Disorder for Social Work with Children. Child and Family Social Work, 5, 291-301.
 Bagley, C., & Pritchard, C. (1998a). The Reduction of Problem Behaviours and School Exclusion in At-Risk Youth: An Experimental Study of School Social Work with Cost-Benefit Analyses. Child and Family Social Work, 3, 219-226.
 Bagley, C., & Pritchard, C. (1998b). The Billion Dollar Cost of Troubled Youth: Prospects for Cost-Effective Prevention and Treatment. International Journal of Adolescence and Youth, 7, 211-225.
 Bagley, C., & Sawyerr, A. (2008). Excluded Youth in Britain: Historical Review of Research and Policy on Inequality, Education, Health, Family Support and Community Development. In C. Bagley, & G. Verma (Eds.), Challenges for Inclusion: Educational and Social Studies from Britain and the Indian Sub-Continent (pp. 183-217). Rotterdam: Sense Educational Books.
 Barnes, J., Belsky, J., Broomfield, K., Dave, S., Frost, M., & Melhuish, E. (2005). Disadvantaged but Different: Variation among Deprived Communities in Relation to Child and Family Well-Being. Journal of Child Psychology and Psychiatry, 40, 952-962.
 Barnes, J. (2007). Targeting Deprived Areas. The Nature of Sure Start Local Programme Neigh-bourhoods. In J. Belsky, J. Barnes, & E. Melhuish (Eds.), The National Evaluation of Sure Start: Does Area Based Intervention Work? (pp. 25-44). Bristol: The Policy Press.
 Belsky, J., Melhuish, E., Barnes, J., Leyland, A., & Romaniuk, H. (2006). Effects of Sure Start Local Programmes on Children and Families: Early Findings from a Quasi-Experimental, Cross-Sectional Study. British Medical Journal, 332, 1476.
 Belsky, J., & Melhuish, E. (2007). Impact of Sure Start Local Programmes on Children and Families. In J. Belsky, J. Barnes, & E. Melhuish (Eds.), The National Evaluation of Sure Start: Does Area-Based Early Intervention Work? Bristol: The Policy Press.
 Briggs, N., Kurtz, A., & Paull, G. (2012). Evaluation of Children’s Centres in England (ECCE) Strand 5: Case Studies on the Costs of Centres in the Most Deprived Areas. London: Department for Education.
 Bywater, T., Hutchings, J., Daley, D., Whitaker, C., Yeo, S., Jones, K., Eames, C., & Edwards, R. (2009). Long-Term Effectiveness of a Parenting Intervention in Sure Start Services in Wales for Children at Risk of Developing Conduct Disorder. British Journal of Psychiatry, 195, 318-324.
 Coe, C., Gibson, A., Spencer, N., & Stuttaford, M. (2008). Sure Start: Voices of the “Hard-to-Reach”. Child Care Health and Development, 34, 447-453.
 Edgley, A., & Avis, M. (2007). The Perceptions of Statutory Service Providers of a Local Sure Start Programme: A Shared Agenda. Health, Social Care and Community, 15, 379-386.
 Edwards, R., ó Céilleachair, A., Bywater, T., Hughes, D., & Hutchings, J. (2007). Parenting Programme for Parents of Children at Risk of Developing Conduct Disorder: Cost-Effective Analysis. British Medical Journal, 334, 682-687.
 Glass, N. (1999). Sure Start: The Development of an Early Development of an Early Intervention Programmes for Young Children in the United Kingdom. Children and Society, 13, 257-264.
 Gray, R., & Francis, E. (2007). The Implications of US Experiences with Early Childhood Interventions for the UK Sure Start Programme. Child Care Health and Development, 33, 655-663.
 Hutchings, J. et al. (2007). Parenting Intervention in Sure Start Services for Children at Risk of Developing Conduct Disorder: Pragmatic Randomized Controlled Trial. British Medical Journal, 334, 678.
 Jones, K., Daley, D., Hutchings, J., Bywater, T., & Eames, C. (2008). Efficacy of the Incredible Years Programme as an Early Intervention for Children with Conduct Problems and ADHD. Child Care Health and Development, 84, 389-390.
 Lewis, J. (2011). From Sure Start to Children’s Centres: An Analysis of Policy Change in English Early Years Programmes. Journal of Social Policy, 40, 71-88.
 Matheny, P. (1995). Bringing Order Out of Chaos: Psychometric Characteristics of the Confusion, Hubbub and Order Scale. Journal of Applied Development Psychology, 16, 429-444.
 Melhuish, E. and the SSLP Study Team (2007). Variation in Community Intervention Programmes and Consequences for Children and Families: The Example of Sure Start Local Programmes. Journal of Child Psychology and Psychiatry, 48, 543-551.
 Melhuish, E. and the SSLP Study Team (2008). Effects of Fully-Established Sure Start Local Programmes on 3-Year-Old Children and Their Families Living in England: A Quasi-Experimental, Observational Study. The Lancet, 373, 1641-1647.
 Melhuish, E., Belsky, J., Leyland, A. et al. (2010). The Impact of Sure Start Local Programmes in Five Year Olds and Their Families: Report of a Longitudinal Study. Research Report DFE-RR067, London: Department for Education.
 Melhuish, E. (2016). Longitudinal Research and Early Years Policy Development in the United Kingdom. International Journal of Child Care and Education Policy, 10, 3.
 Muenning, P., Mohit, B., Wu, J., Jia, H., & Rosen, Z. (2016). Cost-Effectiveness of the Earned Income Tax Credit as a Health Policy Investment. American Journal of Preventative Medicine, 51, 874-881.
 Raymond, J. (2009). “Creating a Safety Net”: Women’s Experiences of Antenatal Depression and Their Identification of Helpful Community Support Services during Pregnancy. Midwifery, 25, 39-49.
 Rutter, M. (2007). Sure Start Local Programmes: An Outsider’s Perspective. In J. Belsky, J. Barnes, & E. Melhuish (Eds.), The National Evaluation of Sure Start. Bristol: The Policy Press.
 Sammons, P., Hall, J., Smees, R. et al. (2015a). Children’s Centres: Their Importance for Children and Families: Evaluation of Children’s Centres in England (ECCE, Strand 4). London: Department for Education. www.gov.uk/government/publications
 Sammons, P., Hall, J., Smees, R., Goff, J. et al. (2015b). Children’s Centres: Their Importance for Children and Families: Evaluation of Children’s Centres in England Summary Report. London: Department for Education. www.gov.uk/government/publications
 Schneider, J., Ramsay, A., & Lowerson, S. (2006). Sure Start Graduates: Predictors of Attainment on Starting School. Child Care Health and Development, 32, 431-440.
 Scott, S., Knapp, M., Henderson, J., & Maughan, B. (2001). Financial Cost of Social Inclusion: Follow-Up Studies of Antisocial Children into Adulthood. British Medical Journal, 323, 191.