Research and Policy in Developing an Early Years’ Initiative: The Case of Sure Start

Abstract

The British government’s wish to eliminate the cycle of disadvantage for children from poor families led to Sure Start. The initiative set up 260 Sure Start Local Programmes (SSLPs) by 2001, which were expanded to 524 programmes within 2 years. SSLPs aimed to enhance the health and development of children under four and their families in deprived communities. SSLPs were area-based, with all children under four and their families in an area being eligible. This allowed efficient delivery of services without stigmatisation. SSLPs did not have a prescribed “protocol” of services. Instead, each SSLP had autonomy to improve and create services, with general goals and some specific targets but without specification of how services were to be delivered. The National Evaluation of Sure Start (NESS) undertook a multifaceted evaluation of SSLPs, and by 2005 research evidence led to a fundamental shift with SSLPs becoming Children’s Centres. The story of how this happened is discussed herein, with latest findings summarised.

A Labour government returned to power in the United Kingdom on May Day 1997, and the notion of “joined-up” services was in vogue. For example, in relation to services for children, the Prime Minister, Tony Blair (1988), commented that “We have looked at key problems across government. The old departmental boundaries often do not work. Provision for young children—health, childcare, support—will be co-ordinated across departments so that when children start school they are ready to learn.” Ministers agreed that there should be a review of services for young children that often appeared to be failing those in greatest need, while there was evidence from programmes in the Unites States like Head Start and the Perry Pre-School programme (Barnett, 1995), as well as from experimental programmes in the UK, that comprehensive early years’ interventions could make a difference to children’s development and longer-term wellbeing.

The “Cross-Departmental Review of Services for Young Children” was to assess whether greater emphasis on preventative action and a more integrated child-centred approach to services could cut crime and unemployment, and reduce the need for extra help for individuals at school and state benefits in later life, by helping parents, carers and communities provide the best possible start for children and thus improve their future lives. The review was designed to: (a) look at the policies and resources for children aged seven and under, to ensure effectiveness in providing preventative action and support to ensure the development of their full potential throughout their lives, (b) consider whether the multiple causes of social exclusion affecting young children could be more effectively tackled at the family and community levels using an integrated approach to service provision, and (c) take account of policy developments elsewhere (HM Treasury, 1998b). This review involved 11 government departments and produced the 1998 Comprehensive Spending Review that made several recommendations, including the following (HM Treasury, 1998b):

• 1. The earliest years in life were the most important for child development, and very early development was much more vulnerable to adverse environmental influences than had previously been realised.

• 2. Multiple disadvantages for young children were a severe and growing problem, with such disadvantage greatly enhancing the chances of social exclusion later in life.

• 3. The quality of service provision for young children and their families varied enormously across localities and districts, with uncoordinated and patchy services being the norm in many areas. Services were particularly dislocated for the under fours—an age group that tended to get missed by other Government programmes.

• 4. The provision of a comprehensive community based programme of early intervention and family support which built on existing services could have positive and persistent effects, not only on child and family development but also in helping break the intergenerational cycle of social exclusion, which could lead to significant long term gain to the Exchequer.

The review argued that the comprehensive community based programme should not be based upon a single blueprint for the ideal early intervention. However programmes should be: (a) two-generational, involving parents as well as children; (b) non-stigmatising, avoiding labeling “problem families”; (c) multifaceted, targeting a number of factors, not just, for example, education or health or parenting; (d) persistent, lasting long enough to make a real difference; (e) locally driven, based on consultation with and involvement of parents and local communities; and (f) culturally appropriate and sensitive to the needs of children and parents. It was argued also that services should be integrated to support the physical, developmental and emotional needs of young children and families. Such services should be easily accessible—within “pram pushing distance”—and backed up by outreach to offer support in the home. The programme was to be area-based, with all children under four and their families living in an area being clients of the programme, with the right to a say in the services provided. This area-based characteristic fitted the model of other area-based initiatives that were a feature of much government policy at that time. There was a general sense that because social welfare problems were often geographically concentrated, that it would be less stigmatising to individuals and families to target areas as being at risk of problems and in need of special attention and extra resources than to target particular at-risk individuals or families.

In an innovation for the development of policy, empirical findings of research studies were taken into account in the review, as described by Norman Glass in oral evidence to House of Commons Science and Technology Committee (2006):

We were influenced very heavily by a series of experimental studies in the United States, many of them different but relating to early years’ programmes, which appeared to show significant improvements on a number of measures. … We were influenced by issues of evidence from our own birth cohort studies which showed that many of the influences in people’s later lives were present in the first seven years of their lives and that those were the most significant influences affecting people’s lives, in so far as you could see what affected people’s lives. There was a lot of evidence on the importance of things like parental attachment and so on. There was a lot of stuff around of that kind which did not point to particular programmes but nevertheless pointed in the direction of saying that early years mattered and probably mattered more than interventions you could make later on in people’s lives and that there were things that appeared to be effective which were being carried out elsewhere. (p. 26)

Evidence from the USA that was influential included randomised control trials of early years’ interventions, demonstrating clear benefits for disadvantaged children of high quality childcare provision, whether started in infancy (Abecedarian Project, Ramey, & Campbell, 1991) or at three years of age (Perry Pre-school Project, Schweinhart et al., 1993). Also, it was noted that where quasi-experimental studies had rigorous methodology, they produced similar results. Small-scale tightly controlled interventions had larger effects than more extensive large-scale interventions, such as the Chicago Child-Parent Centers (Reynolds, Temple, Robertson, & Mann, 2001) and Head Start (Karoly et al., 1998). Nevertheless, the impact of large-scale interventions was still substantial, producing worthwhile benefits for children, families and communities.

Many early childhood interventions used home visiting and parental support as a supplement to childcare, which was said to have additional benefits. There was evidence of the benefits of home visiting provided by nurse-qualified staff, rather than by para-professionals, particularly if a highly structured approach was implemented (Olds et al., 1997; Kitzman et al., 1997). Evidence concerning the importance of the early years and the potential of early intervention (e.g., Shonkoff & Phillips, 2000) has encouraged the development of several intervention projects in the USA, Australia, and Canada, as well as Sure Start Local Programmes (SSLPs) in the UK. All of these programmes set out to influence factors such as parent attitudes, mental health, childrearing, and high-quality early stimulation and education. It is worth noting that the early results from these intervention efforts highlighted the need for caution as only sustained high-quality efforts proved to be effective (Melhuish, 2004; Olds, 2002).

In addition to the lessons from research on early intervention and child health, Norman Glass came to believe in the importance of community involvement and thus advocated a community development approach to early-years intervention. This was somewhat surprising given the nature of the evidence that was used to justify Sure Start spending, which had little, if anything, to say about community development. The emphasis placed on community development in Sure Start was explained by Norman Glass (2005) as follows:

This programme would be “owned” by local parents, local communities and those who worked in the programme. Because those who benefited would be able to shape it to do what they wanted, rather than it being done to, or for, them, it would not be seen as just another initiative by Whitehall. … What I learned from visits to successful early years’ programmes and local communities was that it was necessary, in the case of early years at any rate, to involve local people fully in the development and management of the programme if it was to take root and not simply be seen as another quick fix by middle-class social engineers. (p. 101)

The community development approach of Sure Start was also consistent with broader principles central to New Labour’s interest in modernizing government, including the view that public services should be user not provider driven, evidence based, joined up, and innovative.

The Birth of Sure Start

The findings of the Cross-Departmental review were incorporated into the 1998 Comprehensive Spending Review that delineated future government expenditure. Announcing its details on 14th July 1998, the Chancellor of the Exchequer introduced the plan for Sure Start aiming to bring together quality services for children under four and their parents — nursery, childcare, and playgroup provision, and prenatal and other health services, including an offer to parents of counseling and help to prepare their children for learning and for school (HM Treasury, 1998a).

The review’s final report noted that disadvantage among young children was increasing, that this could result in later difficulties, and that the earlier intervention was undertaken, the more likely it was that poor outcomes could be prevented. Further, the report noted, current services were uncoordinated and patchy, that young children often missed out on services which concentrated on older children, and that the quality of services varied. Nevertheless, there existed good practice that could inform the enhancement of programmes for young children. It was recommended, therefore, that there should be a change of approach to the design and delivery of services. They should be jointly planned by all relevant bodies, both within the local authority and outside it (HM Treasury, 1998b).

A total of £542 million became available over three years, with £452 million designated for England. The four nations of the United Kingdom—England, Wales, Scotland and Northern Ireland—would each develop their own Sure Start plans. In England it was decided that there would be 250 programmes up and running by 2001–02, supporting about 187,000 children, or 18% of all poor children under four. On average, a local programme was to include 800 children under four years of age. The intention was to provide each programme, at the peak of funding, with a budget that could not be later redirected to something else—roughly equivalent to £1,250 per annum per child. Programmes were to be funded for 10 years to ensure money was not diverted to other services and to signal a long-term commitment. Each SSLP was to run for at least seven to 10 years with government funding peaking at year three and declining from year six until reaching zero at year 10. It was implicit that some funding would be picked up by local government authorities and that some funding would no longer be needed because of the successful “reshaping” of mainstream services to more appropriately meet the needs of local families. This commitment and investment utterly transformed early years’ services in the UK, while still representing a relatively small contribution from the perspective of Treasury at just 0.05% of public expenditure.

The SSLP programmes were to be targeted on the most deprived 20% of areas. There were no published figures on how many poor children lived in such areas. However, using statistics from the Indices of Multiple Deprivation (Office of the Deputy Prime Minister, 2004), the National Evaluation of Sure Start (NESS) calculated that for children in families with an income 60% or less than the national median (official poverty line), 51% of all such poor children live in the 20% most deprived areas and 65% live within the 30% most deprived areas.

Joined-up Government and Sure Start

The Sure Start Unit (SSU) responsible for the new initiative was cross-departmental (involving the Departments of Education and Employment; Health; Social Security; Environment, Transport and the Regions; Culture, Media and Sport; Trade and Industry; and the Home Office; Lord Chancellor’s Department and HM Treasuryhoused) and housed in the Department for Education and Skills (since rebranded Department of Children, Schools and Families). Such inter-departmental cooperation could only have been achieved with Treasury’s influence. The embryonic SSU advertised for a Head and Naomi Eisenstadt’s appointment was announced in December, 1998. Also around this time government announcements became more specific about the contents of SSLPs: home visits for all families with newborn children to inform about available services and support; health, education and childcare services; toy libraries; toddler groups and family support.

SSLPs were meant to bring “joined-up” services of health, childcare and play, early education and parental support to families with a child under four years of age. SSLPs were to be a completely new way of working for central and local governments. They were to be the glue that would bind together services for families. They were to be based on the best evidence and on experience of what works to give children and families the very best chance to thrive. Programmes were directed to provide outreach for difficult-to-reach families and could add extra services to suit local needs, such as debt counselling, employment, and benefits advice. Community control was to be exercised through local partnerships. Initially, service-provider organisations in a deprived area were invited to submit a bid for Sure Start funding. The invitation indicated that a partnership of local stakeholders had to be constituted and that this partnership needed a plan for a Sure Start programme, nominating a lead agency. These partnerships were to be at the heart of the initiative and bring together everyone concerned with children, including health, social services, education, the private sector, the voluntary sector and parents. Thus, partnerships were to provide local community influence for the design of each SSLP and, as a consequence, even though core services were required, no specification was provided of how they would be delivered, only what they should aim to achieve. Funding was to flow from central government, the SSU, directly to programmes (i.e., not via local government). Programmes could act largely independently of local government, although local departments of education, social services and the like would typically be part of the partnership.

Central Government Guidance for SSLPs

The Sure Start Unit (1998) prepared guidance for local programmes based on the key principles of the programme. Emergent SSLPs were told that services must co-ordinate, streamline, and improve existing services in the SSLP area, including signposting to existing services; involve parents; avoid stigma; ensure lasting support by linking effectively with services for older children; be culturally appropriate and sensitive to particular needs; be designed to achieve specific objectives relating to Sure Start’s overall objectives; and promote accessibility for all local families, later changed to “promote the participation of all local families in the design and working of the programme” (Sure Start Unit, 1998, p. 12). The first formal and written guidance to programmes also outlined the core services that all SSLPs were expected to provide: (a) outreach and home visiting; (b) support for families and parents; (c) support for good quality play, learning, and childcare experiences for children; (d) primary and community health care and advice about child health and development and family health; and (e) support for people with special needs, including help getting access to specialised services.

Getting Sure Start Started

The speed of funding for SSLPs was to some extent overwhelming, resulting in a somewhat slow start to the establishment of operational status. Only 6% of the millions of pounds allocated in 1999 was spent in that year. Of the 60 local groups invited to submit programme plans in January of 1999, only 15 were approved and allocated funds that year. Another 44 were not approved until June of 2000, after refining their plans. Despite this slow start, and without any information pertaining to the success of the initiative, the Treasury expanded Sure Start—doubling the planned number of programmes from 250 by 2002 to over 500 by 2004, thereby more than doubling expenditure to almost £500 million by 2003–04. The expanded Sure Start initiative was to reach one third of poor children under four years of age. In return for this investment, each SSLP was expected to deliver quantified targets in children’s social and emotional development, health and ability to learn, as well as strengthening families and communities.

This rapid expansion of SSLPs so soon after the initial setting up of the initiative was not universally welcomed and had implications for the evaluation. Some advised that it was too early to double the number of SSLPs, but were ignored, probably because the expanded Sure Start would be too big to be easily cut by any future administration. Thus it was that SSLPs became a cornerstone of the UK Government’s campaign to reduce child poverty and social exclusion. SSLPs were to serve all children under four and their families in a prescribed area. This area-based strategy allowed the relatively efficient delivery of services to those living in deprived areas without stigmatising those individuals and families receiving services: Disadvantaged areas were targeted, but within the area the service was universal.

The Autonomy of SSLPs

As a consequence of the local autonomy central to the community control of SSLPs, they did not have a prescribed “protocol” of services to promote adherence to a prescribed model even though they had a set of core services to deliver. Thus, each SSLP had freedom to improve and create services as they wished, with general goals and some specified targets (e.g., reduce number of low birth weight babies, improve language development of young children), but without specification of exactly how services were to be delivered. This contrasted markedly with interventions with clear models of provision and demonstrable effectiveness that provided evidence justifying Sure Start (e.g., Abecedarian project, Ramey, & Campbell, 1991; Perry Pre-school project, Schweinhart et al., 1993; Incredible Years, Webster-Stratton, 1993). It appeared that while research evidence was critical to winning the argument for increased early years’ expenditure, it was largely overlooked in the actual operation of programmes, despite entreaties that services be ‘evidence based’. Indeed, even though the SSU published guidance offering a menu of ‘evidence-based’ interventions from which to choose, there is little evidence that was used.

Evaluation

One of the conditions insisted upon by Treasury in sponsoring the Sure Start initiative was that there be a rigorous evaluation of the programme. Following competitive tender, NESS was commissioned in early 2001 to undertake a multifaceted evaluation, addressing (a) the nature of the communities in which SSLPs were situated, (b) the ways in which SSLPs were implemented, (c) the impact of SSLPs on children, families and communities and (d) the cost-effectiveness of SSLPs. In addition, NESS was charged with providing technical support to local programmes so that each could undertake its own local evaluation to inform the further service development.

The great diversity amongst SSLPs posed challenges in that there were not several hundred programmes delivering one well-defined intervention, but several hundred unique and multifaceted interventions operating in different places. NESS used a variety of strategies to study the first 260 SSLPs, in particular studying children and families in 150 of these with great intensity. These included the gathering of administrative data already available on the small geographic areas that defined SSLP communities (e.g., census data, police records, work and pension records); developing systems to collate information specific to each SSLP area; conducting surveys of SSLPs dealing with many aspects of SSLPs; carrying out face-to-face and telephone interviews with programme managers, employees, and parents about their local programme; and conducting a large-scale survey of child and family functioning in thousands of households in SSLP areas, and in SSLP-to-be areas. The primary results of NESS up to 2006 are summarised in Belsky, Barnes and Melhuish (2007). We review here some core findings and provide a summary of more recent evidence.

Effects on Children/Families

NESS employed two strategies to evaluate the impact of SSLPs on children and their families. The first involved a wait-list control design, comparing 9-and 36-month old children (and their families) in SSLP areas with age mates growing up in reasonably similar families in communities destined to become SSLP areas. The second involved longitudinally following up a random sub-sample of the 9-month olds in this cross-sectional study when 36-months of age (and 5 years old) and comparing them with children growing up in disadvantaged households who were participating in another large study, the Millennium Cohort Study (MCS). Neither strategy was experimental in nature, because the government proved unwilling to randomly assign communities to the programme. This meant that not only was it necessary to implement multi-level modelling to evaluate SSLP effects, given that children were nested in communities, but that statistical controls needed to be implemented to discount effects of pre-existing differences between SSLP and comparison children, families and communities before testing SSLP effects.

The first results on the impact of SSLPs provided both encouraging and discouraging news to policymakers regarding the anticipated effects of SSLPs (Belsky & Melhuish, 2007; Belsky, Melhuish, Barnes, Leyland, Romaniuk, & the NESS Research Team, 2006; NESS Research Team, 2005). The first, cross-sectional results tapped detectable effects of SSLPs rather early in the programme history, so the ultimate implications would prove conditional and dependent upon the results of the subsequent longitudinal study. This was because children in the longitudinal study would have experienced a greater portion of their lives in SSLP areas at time of outcome measurement than was the case for the 3-year-olds in the cross-sectional study.

Although there was some limited evidence that SSLPs exerted across-the-board effects on some aspects of child and family functioning, the most interesting results indicated that effects appeared to vary by degree of family disadvantage (Belsky et al., 2006; Belsky & Melhuish, 2007). More specifically, three-year-olds of non-teen mothers (86% of sample) exhibited fewer behaviour problems and greater social competence when living in SSLP communities than in comparison communities, and evidence indicated that these effects for children were mediated by SSLP effects on the parenting of non-teens (i.e., SSLP → less negative parenting → better child functioning). Adverse effects of SSLPs emerged for children of teen mothers (14% of sample), however, in terms of lower verbal ability and social competence and more behaviour problems than their counterparts in comparison areas. Children from households in which there was no employed adult (40% of sample) and from lone-parent families (33% of sample) also showed evidence of adverse effects of SSLPs, scoring lower on verbal ability when growing up in SSLP areas than did their counterparts in comparison communities.

These early results suggested that of the children from (mostly) deprived families in deprived communities, those from relatively less (but still) disadvantaged households (i.e., non-teen mothers) in SSLP areas benefited somewhat from living in these areas, perhaps due to the beneficial effects of SSLPs on the parenting of non-teen mothers. In contrast, within these same deprived communities, children from relatively more disadvantaged families (i.e., teen mother, lone parent, workless household) appeared to be adversely affected by living in a SSLP community. In sum, the early findings provided some evidence that relatively less disadvantaged children/ families benefited and relatively more disadvantaged children/families were adversely affected by SSLPs. Although compelling evidence to account for this differential pattern of efficacy did not exist, there was some indication that these differential effects may have resulted from the fact that the relatively less disadvantaged families may have been using more SSLP services. It may even have been the case that the more disadvantaged families felt uncomfortable associating with less disadvantaged families or found the services intrusive.

Programme Variability

The NESS research design afforded the opportunity to illuminate the conditions that might have made some SSLPs more effective than others. Detailed information gathered across several years on each programme (Anning & Ball, 2007; Meadows, 2007; Tunstill & Allnock, 2007) were subjected to systematic quantitative analysis (Melhuish et al., 2007; Melhuish, Belsky, Anning, & Ball, 2007). That is, the mostly qualitative data gathered on 150 SSLPs was quantified to reflect the degree to which each programme successfully realised distinct Sure Start principles. These included offering a range and balance of services, providing quality training for staff, exercising effective leadership and management and having effective strategies for identifying families in the community, to name just several of 18 distinct dimensions of implementation subject to quantitative scoring. Not only did programmes that tended to be rated as high on realising one of these principles tend to score high on the others, making for essentially better and more poorly implemented programmes, but evidence emerged that better implemented programmes yielded somewhat greater benefits. While the evidence was not overwhelming, it was consistent with theory about the conditions under which programmes should prove most effective and provided guidance as to what it takes to generate the kinds of benefits that SSLPs were intended to achieve.

Community-Level Change

Recall that the notion of community development was central to the Sure Start initiative, making it dramatically different from most other early intervention programmes, mostly trialled in the USA, on which it was based. In line with other efforts by government that focused on small geographic areas in attempt to avoid stigmatising at-risk individuals and families while targeting resources where they might generate the most return on investment, Sure Start was area-based, with communities and not just children and/or families as the target of intervention. Ultimately, the view was that children and families could be affected by the programme both directly, via services encountered, and indirectly, via community changes that derived from the programme (e.g., reductions in crime, feelings of cohesion, changed ‘local norms’ about parenting).

Reflecting this focus on community change, NESS set out to document characteristics of Sure Start areas over time and examined the relationship between changes in these and programme operations. Community characteristics were tracked over five years, drawing on a wide variety of data (Barnes, 2007a, 2007b; Barnes et al., 2007). It was possible to link these data with Sure Start areas using geographic information system strategies (Frost & Harper, 2007). Community changes were chronicled from January 2000 to May 2004 and compared with changes taking place over all of England over the same period (Barnes, 2007b; Barnes et al., 2007). It would have been preferable to compare SSLP areas with similarly disadvantaged neighbourhoods that did not have SSLPs, but annual information on most community indicators was not available in sufficient detail. Nevertheless, the statistical comparison with change in England proved instructive.

Over a five-year period some community-level improvements in SSLP areas were detected, though many mirrored trends in England, making it difficult to conclude that these changes were a result of SSLPs per se, and few could be linked to programme characteristics such as time in operation. However, some changes were associated with other area characteristics, either the existence of other government sponsored ‘area-based initiatives’, or the extent of deprivation, or the proportion of minority ethnic groups, or variability in housing or deprivation.

SSLP areas became home to more young children over time while the proportion living in households totally dependent on benefits, or in receipt of benefits indicating a job seeker or someone on a low wage, decreased markedly. For instance, the proportion of children under 4 living in workless households in SSLP areas dipped below 40%, having started out at 45% in January 2000. On average, 33% were living in a household in receipt of government Income Support, down from 39%. These average levels were still much higher than the rates of England overall (22% and 18%), but revealed important improvements though there was still vast variability across SSLP areas (range 12% to 64%).

Some aspects of crime and disorder also improved in SSLP areas, notably burglary and exclusions/suspensions of children from school, as well as unauthorised absences by children from schools (i.e., truancy). Moreover, children from age 11 upwards demonstrated improved academic achievement, particularly when there were other government-sponsored area-based initiatives operating locally. While infant health did not improve, reductions in emergency hospitalisations of young children aged 0 to 3 for severe injury and for lower respiratory infection indicated that families in SSLP areas may have been accessing routine health care, at local doctors’ offices or child health clinics, supported possibly by more ‘joined-up’ working between health and social services. It appeared that health screening of young children was enhanced over five years in SSLP areas, and children identified with special educational needs or eligible for benefits related to disability increased significantly.

Subsequent Policy Developments

The need for greater coordination of children’s services led, in 2003, to the creation of a Minister for Children, Young People and Families, and policy-making did not stand still while the SSLP evaluation was underway. When the initial and early NESS evidence of impact upon children and families became available, it contributed to fundamental change in SSLPs. The NESS findings indicated that SSLPs were not having the impact hoped for. Also, evidence from another ongoing research project, the Effective Provision of Pre-school Education (EPPE) (see Sylva, Melhuish, Sammons, Siraj-Blatchford, & Taggart, 2004), showed that a particular type of early years’ provision, integrated Children’s Centres, was particularly beneficial to children’s development. Margaret Hodge, as government Minister for Children Young People and Families, was responsible for Sure Start and she decided that this combination of evidence justified transforming SSLPs into Children’s Centres. This was announced in 2005 alongside the transfer of Sure Start Children’s Centres into Local Authority control. This transfer of control from central to local government was politically inspired to ensure that Sure Start Children’s Centres became embedded within the welfare state by government statute and would thus be difficult to eradicate by any future government. Nevertheless, the transfer of control to local authorities proved unpopular with many Sure Start advocates.

These changes meant that from April 2006, local authorities became accountable for Sure Start Children’s Centres (rather than community partnership boards), and health agencies were legally obliged to cooperate in providing services within Children’s Centres. The money allocated for Children’s Centres and associated programmes were £1.3 billion in 2005–06. For 2006–07, £1.7 billion was provided for Children’s Centres. For 2007–08, £1.8 billion was set aside, representing almost four times the amount spent on equivalent services in 2001–02. Sure Start thus became a significant part of the Welfare State. As Prime Minister Blair (2006) stated:

Sure Start is one of the government’s greatest achievements. It is a programme that gives antenatal advice, and early-years help for children who need it. It is a vital source of learning to parents who often find work on the back of it; and a community facility that becomes a focal point for local health, childcare and educational networks. It has become a new frontier of a changing welfare state. (p. 2)

Changes in Sure Start and Changing Evidence of Impact

The NESS Impact Study team carried on with the longitudinal investigation of children and families earlier seen at 9 months of age, being seen again at 3 years of age. After taking into consideration preexisting family and area characteristics, comparisons of children and families living in SSLP areas with those living in similar areas not receiving SSLPs (and enrolled in the MCS) revealed a variety of beneficial effects for children and families living in SSLP areas, when children were 3 years old (NESS, 2008; Melhuish, Belsky, Barnes, Leyland and the NESS Research Team, in press). There were positive effects associated with Sure Start programmes for 7 of the 14 outcomes assessed. Children in Sure Start areas showed better social development, exhibiting more positive social behaviour, and greater independence/self-regulation than their counterparts not living in Sure Start areas. Parenting, too, was positively affected, with families in Sure Start areas showing less negative parenting while providing their children with a better home learning environment. The beneficial parenting effects appeared to be responsible for the higher level of positive social behaviour in children in Sure Start areas (i.e., SSLP → less negative parenting → enhanced social development). Also families in Sure Start areas reported using more services designed to support child and family development than did families not in Sure Start areas.

The results of this second phase of impact evaluation differed markedly from the first phase (Belsky et al., 2006; Belsky & Melhuish, 2007; NESS 2005). Whereas earlier findings indicated that the most disadvantaged 3-year-old children and their families (i.e., teen parents, lone parents, workless households) were doing less well in Sure Start areas, while somewhat less disadvantaged children and families benefited (i.e., non-teen parents, dual parent families, working households), the most recent findings of the impact evaluation provided almost no evidence of adverse effects of Sure Start programmes. Indeed, the Sure Start effects appeared generalisable across population sub-groups (e.g., workless households, teen mothers) for two reasons: (a) In general, there were almost no consistent differences in effects of Sure Start programmes for particular subgroups and, (b) there was almost no consistent evidence that children and families in the most disadvantaged Sure Start areas, which had more of the most disadvantaged families, functioned more poorly than children and families in somewhat less disadvantaged Sure Start areas.

Various explanations can be offered for the dramatic difference in results between the earlier 2005 findings and the current results. Differences could have arisen because of methodological differences. Although there is no way to determine whether this was the case, it seems eminently possible that the contrasting results accurately reflect the contrasting experiences of SSLP children and families participating in the two phases of the NESS Impact Study evaluation. Whereas those three-year-olds enrolled in the first phase were exposed to relatively immature programmes—and probably not for their entire lives—the three-year-old children and their families participating in the second phase were exposed to more mature and better developed programmes throughout the entire lives of the children. Also, these latter children and families were exposed to programmes that had the opportunity to learn from the results of the first phase of evaluation, especially with respect to the need for greater effort to be made to reach the most vulnerable households. In sum, differences in the amount of exposure to these programmes and the quality of Sure Start programmes may well account for both why the first phase of impact evaluation revealed some adverse effects of the programme for the most disadvantaged children and families and why the second phase of evaluation revealed beneficial effects for almost all children and families living in Sure Start areas.

Conclusion

Sure Start Local Programmes were part of a revolution in services initiated by the New Labour government when it came to power in 1997. While putatively based on evidence from well known American early intervention, it adopted a strikingly different model in attempt to break the intergenerational cycle of poverty by enhancing the life prospects of young children under four growing up in disadvantaged families in disadvantaged communities. Rather than providing a narrowly defined parent-training home-visiting service or a quality-child-care service, it sought to enhance numerous facets of communities and in ways that would not stigmatise children and families in need. Time—and the NESS—has shown that the broad, ambitious, but ultimately unfocussed remit of the original SSLP model was less effective than anticipated. Wisely, instead of “shooting the messenger”—that is, the independent evaluators— the government took on board the findings from the first phase of impact evaluation, modified the programme, focusing more on quality child care and parent support via Children’s Centres, and the change seems to be yielding benefits. Ultimately, though, in the absence of a random-control trial which would have involved randomly assigning communities to the programme, even in a delayed, but systematic way, it will always remain difficult to be certain that the apparent effects of SSLPs on child and family functioning were truly a result of the programmes themselves rather than some unmeasured and confounding factors that went uncontrolled in the evaluation. For the same reason, it is impossible to know whether the changes chronicled at the community level reflect change caused by the community-development ethos of SSLPs or simply change taking place in the larger society.

One lesson, therefore, from the evaluation of Sure Start is that it is always best to adopt a more experimental approach and one with an intervention clearly defined, or even manualised, rather than one that articulates philosophy and goals mostly, and far less so the means for achieving them. The other, though, is that a commitment to enhance child and family well being should not lead to an excessive devotion to any particular strategy, no matter how promising it seems to be or even how much it is heralded by advocates. Being ready to change in response to evidence is the best way to realise the original commitment in the first place. No government should expect to solve the major problems that Sure Start sought to address in a first attempt. Being ever ready to revise, learning even hard lessons, should always be the order of the day.