Content uploaded by Edward Melhuish
Author content
All content in this area was uploaded by Edward Melhuish on Jan 28, 2016
Content may be subject to copyright.
613318
CARE
Curriculum Quality Analysis and Impact Review of European Early Childhood
Education and Care (ECEC)
Instrument: Collaborative project
Call Identifier: FP7-SSH-2013-2
Early childhood education and care: promoting quality for individual, social and economic benefits
D4.1: A review of research on the effects of Early Childhood Education and Care
(ECEC) upon child development
This report considers international research on the impact of Early Childhood Education and Care (ECEC)
provision upon children’s development, using studies reported from a wide range of sources including journals,
books, government reports and diverse organisation reports.
High-quality childcare has been associated with benefits for children’s development, with the strongest effects
for children from disadvantaged backgrounds. There is also evidence that negative effects can sometimes occur.
The results of studies partly depend upon the context and ECEC systems in place in different countries, but there
is sufficient commonality of findings to indicate that many results are not culture-specific.
Discrepant results may relate to age of starting and also differences in the quality of childcare. In addition,
childcare effects are moderated by family background with negative, neutral and positive effects occurring
depending on the relative balance of quality of care at home and in childcare. Recent large-scale studies find
effects related to both quantity and quality of childcare. The effect sizes for childcare factors are about half those
for family factors. The analysis strategy of most studies attributes variance to childcare factors only after family
factors has been considered, and, where the two covary, this can produce conservative estimates of childcare
effects.
DUE DATE OF DELIVERABLE: 31.12.2014
ACTUAL SUBMISSION DATE: 31.12.2014, 14.9.2015
Start date of project: 01-01-2014 Duration: 36 Months
CARE contractor:
Utrecht University
613318 Deliverable 4.1
CARE Project Review of research on impact of ECEC
CARE: http://ecec-care.org/ - page 2
A review of research on the effects of Early Childhood Education and Care (ECEC) upon child development
Organisation: University of Oxford
Authors (main authors in bold): Edward Melhuish, Katharina Ereky-Stevens, Konstantinos Petrogiannis, Anamaria
Ariescu, Efthymia Penderi, Konstantina Rentzou, Alice Tawell, Pauline Slot, Martine Broekhuizen, Paul Leseman
Email: edward.melhuish@education.ox.ac.uk
Number of PM: 11.50
Dissemination Level: PU
S U G G E S T E D C I T A T I O N :
Melhuish, E., Ereky-Stevens, K., Petrogiannis, K., Ariescu, A., Penderi, E., Rentzou, K., Tawell, A., Slot, P.,
Broekhuizen, M., & Leseman, P. (2015). A review of research on the effects of early childhood Education and
Care (ECEC) upon child development. CARE project; Curriculum Quality Analysis and Impact Review of
European Early Childhood Education and Care (ECEC). Available at http://ecec-care.org/resources/publications/
Project co-funded by the European Commission within the Seventh Framework Programme (2014-2017)
Dissemination Level
PU
Public
PP
Restricted to other programme participants (including the Commission Services)
RE
Restricted to a group specified by the consortium (including the Commission Services)
CO
Confidential, only for members of the consortium (including the Commission Services)
613318 Deliverable 4.1
CARE Project Review of research on impact of ECEC
page 3 CARE: http://ecec-care.org/
ACKNOWLEDGEMENTS
We are grateful to the European Commission for funding the CARE project (Curriculum Quality Analysis and
Impact Review of European ECEC), and to all colleagues on the research consortium for contributing to this
report. We would also like to thank all members of the advisory group for their participation in discussions, as
well as their very valuable feedback.
613318 Deliverable 4.1
CARE Project Review of research on impact of ECEC
CARE: http://ecec-care.org/ - page 4
Executive Summary
This report considers international research on the impact of Early Childhood Education and Care (ECEC)
provision upon children’s development and, while not exhaustive, is an extremely comprehensive review, using
studies reported from a wide range of sources including journals, books, government reports and diverse
organisation reports.
Early research was primarily concerned with whether children attending non-parental care developed differently
from those not receiving such care. Later work recognised that childcare is not unitary and that the quality or
characteristics of experience matters. Further research drew attention to the importance of the interaction
between home and out of home experience. High-quality childcare has been associated with benefits for
children’s development, with the strongest effects for children from disadvantaged backgrounds. There is also
evidence that negative effects can sometimes occur. The results of studies partly depend upon the context and
ECEC systems in place in different countries, but there is sufficient commonality of findings to indicate that many
results are not culture-specific.
While the research on preschool education (aged three and above) is fairly consistent, the research evidence on
the effects of childcare (birth to three years) has been more equivocal with some negative effects, some null
effects and some positive effects. Discrepant results may relate to age of starting and also differences in the
quality of childcare. In addition, childcare effects are moderated by family background with negative, neutral and
positive effects occurring depending on the relative balance of quality of care at home and in childcare. Recent
large-scale studies find effects related to both quantity and quality of childcare. The effect sizes for childcare
factors are about half those for family factors. The analysis strategy of most studies attributes variance to
childcare factors only after family factors has been considered, and, where the two covary, this can produce
conservative estimates of childcare effects.
Summary of evidence for disadvantaged children
The evidence on ECEC in the first three years for disadvantaged children indicates that high-quality ECEC can
produce benefits for cognitive, language and social development. Low-quality childcare produces either no
benefit or negative effects. High-quality childcare with associated home visits appears to be an effective package
of services.
With regard to provision for three years onwards, disadvantaged children benefit particularly from high-quality
preschool provision. Also children benefit more in socially mixed groups rather than in homogeneously
disadvantaged groups. A number of interventions have shown improvements in cognitive development, but in
some cases such benefits have not persisted throughout children’s school careers. This appears to be partly
because subsequent poor school experiences for disadvantaged children overcome earlier benefits from high-
quality ECEC experience. However early childhood interventions do boost children’s confidence and social skills,
which provides a better foundation for success at school (and subsequently in the workplace). Reviews of the
research often infer that it is the social skills and improved motivation that lead to lower levels of special
education and school failure and higher educational achievement in children exposed to early childhood
development programs. However there is clear evidence that cognitive, language and academic skills can also be
enhanced by ECEC experience and these are likely to play a role in the later educational, social and economic
success that is often found for well-implemented ECEC interventions. Studies into adulthood indicate that this
educational success is followed by increased success in employment, social integration and sometimes reduced
criminality. There is also an indication of improved outcomes for mothers. The greatest improvements appear to
occur for those problems that are endemic for the particular disadvantaged group, where there is greatest
opportunity for improvement, e.g. behaviour problems, criminality and lack of educational achievement.
613318 Deliverable 4.1
CARE Project Review of research on impact of ECEC
page 5 CARE: http://ecec-care.org/
Summary of evidence for the general population
The evidence on ECEC in the first three years indicates that for children who are not disadvantaged in their home
environment, high-quality ECEC benefits children’s cognitive, language and social development in both the short-
and long-term, but low-quality childcare can be a risk factor and may lead to a dual risk for children from low-
income families, leading to possible deficits in language or cognitive development. There has been some
evidence that high levels of childcare, particularly group care in the first two years, may elevate the risk for
developing antisocial behaviour. However subsequent research indicates that this may be related to high levels of
poor quality care, particularly in centres and in the first year.
The low level of much ECEC quality is of concern. Some have argued (e.g. see Haskins & Barnett, 2010) that, in
the US for example, government-funded preschool programs (e.g. childcare centres, Head Start and state-funded
prekindergarten) offer services that are of ‘mediocre or worse’ quality, that children attending the average
program may gain little cognitive boost, and that greater benefits could be gained by improving the quality of
these programs. Others (e.g. Pianta, Barnett, Burchinal, & Thornburg, 2009) maintain that publicly funded
preschool in the US narrows the achievement gap between poor and non-poor groups by as little as five per cent
because of the prevalence of low-quality programs and that preschool could be narrowing the gap by up to 50
per cent if quality were improved. Furthermore, it has been argued that, while ECEC for children at risk can
contribute importantly to combating educational disadvantage, this can only occur if certain circumstances are
met. The design of programs and pedagogy and curriculum are crucial (Leseman, 2009).
For provision for three years onwards the evidence is consistent that preschool provision is beneficial to
educational and social development for the whole population. An example of the multi-national nature of
positive ECEC effects is provided by an (OECD, 2011) report on PISA results that found that students who had
attended some pre-primary school outperformed students who had not, by about a year of achievement. Studies
indicate that the benefits are greater for high-quality provision. Some evidence shows that part-time provision
produces equivalent effects to full-time provision for the general population but more deprived children may
benefit from full-time. Also there is evidence that a starting age from two years of age onwards is most effective
for preschool education.
Characteristics of Early Years provision and child development
Children’s daily experiences drive child development. To optimise the child experience it is possible to act upon
several characteristics and hence improve the quality of ECEC. Aggregating evidence across much research
indicates that the following quality characteristics of Early Years provision are important for enhancing children's
development:
1. Adult-child interaction that is responsive, affectionate and readily available
2. Well-trained staff who are committed to their work with children
3. A developmentally appropriate curriculum with educational content
4. Ratios and group sizes that allow staff to interact appropriately with children
5. Supervision that maintains consistency in the quality of care
6. Staff development that ensures continuity, stability and improving quality
7. Facilities that are safe and sanitary and accessible to parents
To promote stronger outcomes, ECEC should be characterized by both structural features of quality and ongoing
supports to teachers to assure that the immediate experiences of children, those provided through activities and
interactions, are rich in content and stimulation, while also being emotionally supportive, and adapted to the
children’s developmental level. In addition to in-classroom professional development supports, the pre-service
training and education of ECEC staff is of critical concern. However, here evaluation research is still scant. There
are a range of recent innovations, however, these innovations have yet to be fully evaluated for their impact on
staff capacities or ECEC quality.
613318 Deliverable 4.1
CARE Project Review of research on impact of ECEC
CARE: http://ecec-care.org/ - page 6
Complex pathways in child development
Child development is affected by the whole range of children’s experiences, particularly in the Early Years, and
ECEC is a substantial part of the young child’s experience. Also as children enter school, experiences in that
environment will influence longer-term outcomes. Not only do ECEC experiences play an important role in
promoting child well-being, but some other background factors are also important. The relevant factors do not
function alone, but interact with each other. Hence the potential effects of ECEC experience are partly
moderated by family factors, such as deprivation and parental sensitivity as well as child factors such as gender,
temperamental reactivity and self-regulation. Sometimes the moderating variable may itself be influenced by
ECEC experience, e.g. self-regulation, and when this occurs the distinction between moderating and mediating
variable becomes blurred. In the case of self-regulation it appears to be important in the process by which early
family and ECEC experiences get transmitted into later educational, social and economic success.
613318 Deliverable 4.1
CARE Project Review of research on impact of ECEC
page 7 CARE: http://ecec-care.org/
Table of Contents
Introduction ................................................................................................................................................................... 10
A historical perspective of ECEC research ................................................................................................................... 11
Part 1: ECEC as an intervention for children from disadvantaged backgrounds .......................................................... 12
US studies of ECEC as an intervention for children from disadvantaged backgrounds ............................................ 13
ECEC for children aged zero to three years ............................................................................................................ 13
Early Head Start (EHS) ......................................................................................................................................... 13
Infant Health and Development Program (IHDP) .............................................................................................. 14
Milwaukee Project ............................................................................................................................................... 15
Abecedarian Project ............................................................................................................................................ 15
Project CARE ........................................................................................................................................................ 16
ECEC for children aged over three years ................................................................................................................ 17
Randomised Control Trials (RCTs) ...................................................................................................................... 17
Perry Preschool Project (PPP) ........................................................................................................................ 17
Early Training Project (ETP) ............................................................................................................................ 18
Head Start........................................................................................................................................................ 19
Quasi-experimental evaluations of Head Start ...................................................................................................... 22
Cognitive outcomes ............................................................................................................................................. 22
Non-cognitive outcomes ..................................................................................................................................... 22
Quasi-experimental studies of other US programs ................................................................................................ 22
Child-Parent Centers (CPC) ................................................................................................................................. 22
Miami school readiness program ....................................................................................................................... 24
Great Start Readiness Program .......................................................................................................................... 24
Texas Targeted Pre-Kindergarten Program ........................................................................................................ 25
Syracuse Family Development Research Program ............................................................................................ 26
Delaware Early Childhood Longitudinal Study ................................................................................................... 26
European studies of ECEC as an intervention for children from disadvantaged backgrounds ................................. 27
Randomised Control Trials (RCTs) ........................................................................................................................... 27
UK ......................................................................................................................................................................... 27
Denmark ............................................................................................................................................................... 28
Quasi-experimental and longitudinal studies ......................................................................................................... 28
Germany............................................................................................................................................................... 28
Netherlands ......................................................................................................................................................... 29
France ................................................................................................................................................................... 30
Summary of evidence for disadvantaged children ...................................................................................................... 30
Part 2: ECEC for the General Population ....................................................................................................................... 31
Attendance patterns of care ........................................................................................................................................ 31
Attendance and quantity of childcare ......................................................................................................................... 31
613318 Deliverable 4.1
CARE Project Review of research on impact of ECEC
CARE: http://ecec-care.org/ - page 8
ECEC for children aged zero to three years ............................................................................................................ 31
Socio-emotional development ........................................................................................................................... 32
Attachment ..................................................................................................................................................... 32
Other aspects of Social and Emotional Development .................................................................................. 33
Cognitive, language and educational development ...................................................................................... 35
US Studies ................................................................................................................................................... 35
Australasian studies ................................................................................................................................... 36
European and international studies .......................................................................................................... 37
ECEC for children aged over three years ................................................................................................................ 37
Socio-emotional development ........................................................................................................................... 37
Cognitive, language and educational development .......................................................................................... 38
US studies ........................................................................................................................................................ 38
European and international studies............................................................................................................... 40
Developing countries ...................................................................................................................................... 42
The search for the ideal starting age, duration and intensity of attendance ............................................................ 43
Duration .................................................................................................................................................................... 44
Intensity .................................................................................................................................................................... 45
Centre-based versus home-based attendance of care ............................................................................................... 46
ECEC for children aged zero to three years ............................................................................................................ 46
ECEC for children aged over three years ................................................................................................................ 48
Variation in the quality of ECEC.................................................................................................................................... 49
ECEC for children aged zero to three years ............................................................................................................ 51
ECEC for children aged over three years ................................................................................................................ 51
Positive relationships and interactions between practitioners and children ............................................................ 53
ECEC for children aged zero to three years ............................................................................................................ 53
ECEC for children aged over three years ................................................................................................................ 55
Pedagogical practices and curriculum ......................................................................................................................... 56
ECEC for children aged zero to three years ............................................................................................................ 56
ECEC for children aged over three years ................................................................................................................ 57
Developmentally Appropriate Practice (DAP) versus didactic instruction ............................................................ 60
Stability and continuity of care .................................................................................................................................... 62
The physical environment ............................................................................................................................................ 63
Adult-child ratios and group sizes ................................................................................................................................ 65
ECEC for children aged zero to three years ............................................................................................................ 66
ECEC for children aged over three years ................................................................................................................ 66
Practitioner qualifications and training ....................................................................................................................... 67
ECEC for children aged zero to three years ............................................................................................................ 68
ECEC for children aged over three years ................................................................................................................ 69
Part 3:Complex pathways from ECEC to child outcomes ............................................................................................. 71
613318 Deliverable 4.1
CARE Project Review of research on impact of ECEC
page 9 CARE: http://ecec-care.org/
Factors moderating ECEC effects across ECEC age groups .................................................................................... 71
Subsequent educational experience .................................................................................................................. 71
Family Demographic factors and ECEC effects overall ...................................................................................... 72
Child factors and ECEC effects overall ................................................................................................................ 73
The interaction of attendance of ECEC with family and child characteristics, by age groups ............................. 74
ECEC for children aged zero to three years ....................................................................................................... 74
Family background as moderator .................................................................................................................. 74
Child characteristics as moderator, by age groups ....................................................................................... 75
ECEC for children aged over three years ............................................................................................................ 75
Family background as moderator .................................................................................................................. 75
Child characteristics as moderator ................................................................................................................ 77
The interaction of quality of ECEC with family and child characteristics, by age groups..................................... 77
ECEC for children aged zero to three years ....................................................................................................... 77
Family background as moderator .................................................................................................................. 77
Child characteristics as moderator ................................................................................................................ 79
ECEC for children aged over three years ............................................................................................................ 79
Family background as moderator .................................................................................................................. 79
Child characteristics as moderator ................................................................................................................ 80
Conclusion ..................................................................................................................................................................... 81
Summary of evidence for disadvantaged children ................................................................................................. 82
Summary of evidence for the general population ................................................................................................. 82
Characteristics of Early Years provision and child development ........................................................................... 83
Complex pathways in child development ............................................................................................................... 83
Policy relevance ....................................................................................................................................................... 84
References ..................................................................................................................................................................... 85
613318 Deliverable 4.1
CARE Project Review of research on impact of ECEC
CARE: http://ecec-care.org/ - page 10
I N T R O D U C T I O N
Internationally, the number of children attending non-parental childcare and education services before primary
school entry has been increasing since the 1960s, and in developed countries some preschool education or care
is the norm for most groups of children.
‘Today’s rising generation in the countries of the OECD is the first in which a majority are spending a
large part of their early childhoods not in their own families but in some form of childcare’ (UNICEF
Innocenti Research Centre, 2008:3).
The terms day care, child care and Early Childhood Education and Care (ECEC) have all been used to refer to
various forms of non-parental childcare and early education occurring before school. This could include relatives
(e.g. grandmothers), family day care, and group or centre-based childcare and early education. Sometimes, such
ECEC has an explicit educational component (e.g. preschool centres) and sometimes not. However, in that all
experience can potentially be educational, this distinction is not clear-cut. The vast literature on ECEC spans the
disciplines of developmental psychology, education, economics, and beyond, with interest from researchers,
practitioners, policy makers, and the general public.
ECEC has become a salient developmental context for most children in high-income countries, and increasingly
so in low- and middle-income countries. Also, ECEC provision has implications for fertility rates and female
workforce participation, and is consequently embedded in a broader context of educational and family policies.
Rates of ECEC use, types of ECEC, and the content and quality of ECEC differ by child age and socio-political
context. For instance, on average across OECD countries, 67 per cent of three year-olds, and 94 per cent of five
year-olds were enrolled in paid ECEC of some form in 2011 (see www.oecd.org). For children under three,
amongst OECD countries, the use of ECEC varies greatly, from ten per cent and lower in some countries (e.g.
Czech Republic and Poland) to around 60 per cent in Scandinavian countries, with the OECD average being 33 per
cent.
Countries vary considerably in ECEC and parental leave policies. Some countries, (e.g. Sweden and Norway) offer
universal access to subsidized and quality regulated ECEC from infancy, following extensive parental leave. In
contrast, the US has limited support for ECEC or parental leave, and public funding for ECEC is targeted at low-
income preschool children in most states, although an increasing number of states now offer universal preschool
from age three or four. Due to limited availability of parental leave in some countries (e.g. in the US), many
children enter ECEC early during their first year. Moreover, ECEC constitutes a considerable financial burden on
families; across the OECD, ECEC costs 12 per cent of an average family’s income, with the UK and Switzerland
being the most expensive for families with 27 per cent and 50 per cent, respectively (www.oecd.org).
613318 Deliverable 4.1
CARE Project Review of research on impact of ECEC
page 11 CARE: http://ecec-care.org/
A H I S T O R I C A L P E R S P E C T I V E O F E C E C R E S E A R C H
Research on ECEC and child development goes back more than five decades. The first wave of research focused
on comparisons between children in non-parental day care or at home, and grew out of attachment research and
concerns about consequences of early separations from the mother. There were methodological limitations, and
variation in day care settings and differences among children or families were not adequately addressed. The
second wave of research included diverse populations and recognised the importance of measuring variations in
the quality of ECEC for understanding influences upon the development of children. The third wave of ECEC
research took a social-ecological approach (Bronfenbrenner, 1989), in which both family context (especially
family disadvantage and poverty), as well as child characteristics (e.g. temperament), were considered to interact
with the characteristics of ECEC in its influence on child development. Researchers started acknowledging that
children from more advantaged families often receive higher quality ECEC than children from less advantaged
families, unless disadvantaged children attend subsidized ECEC which is used as an intervention to improve
developmental outcomes.
The earliest ECEC research was particularly interested in attachment security and parenting, with interest being
maintained in socio-emotional outcomes (especially externalizing and internalizing behaviours and social skills),
and increasingly in cognitive and academic outcomes (including language development and school achievement).
Moreover, there has also been increasing interest in stress responses (measured via change in cortisol levels) to
caregiving environments, as outcomes of day care experiences.
Even though ECEC research has evolved, many of the same research questions persist, especially concerning
quantity and quality of care. Quantity issues concern whether child development is related to (a) use of non-
parental day care versus parental care, or the use of different types of care; (b) the age at which children enter
ECEC; and (c) the amount of time (e.g. hours per week, six hours in one setting or three plus three hours in two
settings) children spend in ECEC. Quality of ECEC can be considered in terms of structural and process quality.
Structural quality comprises the organisational and physical features of ECEC, and is in general considered higher
when child group sizes and child- adult ratios are small, teachers are trained, and curriculum/program type, toys
and learning materials, and physical space are age-appropriate and adequate. Process quality refers to the quality
of the children’s daily experiences, including adult-child interactions, that foster children’s development, and is
often measured with systematic observational rating scales. Although structural and process quality are
correlated, the relationship may be strong or weak, e.g. sensitive and stimulating interactions may occasionally
take place when staff training is poor or child-adult ratio is high. Nevertheless, process quality is more proximal to
the child, and therefore more directly relevant to developmental outcomes.
While many questions in ECEC research, as well as the outcomes studied, have persisted, recent ECEC research
has evolved along three lines. The first involves context: while much research has been conducted in the US and
the UK, studies from other countries with other ideologies, policies and practices have become increasingly
common, including studies from low and middle-income countries such as Chile and Bangladesh (see Engle et al.,
2011; Love et al., 2003 for extended discussions). The second concerns time span: studies have now followed
children into school age, and even adulthood. The third is methodological, particularly selection effects:
researchers are increasingly aware that parent choices have a strong influence on the ‘when and what’ of
children’s ECEC experiences, and that apparent effects of ECEC on children’s development may, in part, be a
function of family characteristics that vary with selection into ECEC. Studies of the effects of ECEC on child
outcomes may therefore be biased if family factors both influencing ECEC variables and child outcomes are not
controlled for, poorly measured, or not measured. Hence, the better research studies will include measurements
of substantial family characteristics. Although a few studies of ECEC interventions for disadvantaged children
have used randomized controlled trials (RCTs), most ECEC research is based on observational studies, because
having researchers control families’ ECEC choices is unacceptable to most families.
This review deals with research on the possible influences of ECEC on children’s development, including
attachment security, socio-emotional development, and cognitive and academic development. It uses empirical
studies primarily, and literature reviews when relevant. Note that when the term effect is used it is in a broad
sense, referring to both associations and to plausibly causal relationships.
613318 Deliverable 4.1
CARE Project Review of research on impact of ECEC
CARE: http://ecec-care.org/ - page 12
P A R T 1 : E C E C A S A N I N T E R V E N T I O N F O R C H I L D R E N
F R O M D I S A D V A N T A G E D B A C K G R O U N D S
Childcare or preschool education has been used as an intervention strategy to improve the lives and
development of specific groups, particularly children living in deprived circumstances. Children from
disadvantaged/impoverished family backgrounds often experience particular difficulties at school. They enter
school with fewer academic skills than their more advantaged peers, and they often lag behind in their cognitive
development during the later school years (Stipek & Ryan, 1997). More than 40 years of research have convinced
scholars and policy makers that quality preschool experiences benefit children from impoverished environments
and help prepare them for school entry (see, for example, reviews by Barnett, 1995; Brooks-Gunn, 2003;
Committee on Developments in the Science of Learning et al., 2000; Frede, 1995; Haskins, 1989; Heckman, 2006;
Melhuish, 2004b; National Research Council, 2001; Yoshikawa, 1995).
Several studies investigate the effects of these intervention programs on the development of children from
disadvantaged backgrounds. In some cases, the degree of control over the intervention and the potential
recipient population has allowed an intervention based on a RCT procedure to be adopted. In a RCT, assignment
to intervention or control groups is random, hence theoretically balancing groups on background factors that
may influence the results. Where properly executed this is the most powerful evaluation strategy. For more
widespread interventions this level of control usually has not been possible and evaluations have adopted quasi-
experimental designs where group assignment is not randomized, and control for background factors is carried
out by statistical adjustment. This strategy has a potential flaw in that there may be an unacknowledged
background factor that may affect the results. However such designs do allow interventions to be assessed in
typical or usual circumstances and hence produce results of potentially greater generalisability.
613318 Deliverable 4.1
CARE Project Review of research on impact of ECEC
page 13 CARE: http://ecec-care.org/
U S S T U D I E S O F E C E C A S A N I N T E R V E N T I O N F O R
C H I L D R E N F R O M D I S A D V A N T A G E D B A C K G R O U N DS
ECEC FOR C HILD REN AG ED Z ERO TO THR EE YEARS
Several US programs have used ECEC from zero to three years as an intervention for children from disadvantaged
backgrounds. The evaluations of programs were all RCTs and include:
Early Head Start
Infant Health and Development Program
Milwaukee Project
Abecedarian Program (Carolina)
Project CARE
EA RLY HEA D START (E H S)
EHS is a two-generation intervention program serving parents and children from birth to age three, targeted
within disadvantaged communities. It began in 1995 and by 2003 had grown to over 700 programs serving more
than 62,000 children in the US. EHS aims to promote children's development and provides childcare,
developmental assessments, health and parenting services. There are three models of intervention; centre-
based, home visiting, and a combination of these two.
Evaluation of EHS has included a RCT involving 17 EHS sites and following 3,000 EHS children and controls (Love
et al., 2002). Mixed results have been found depending on the cohort of children. For the cohort of three year-
olds, there have been positive effects and for the four years old cohort null or negative effects have been found.
The positive effects for children of EHS participation include:
• Better cognitive and language development;
• better immunisation records and less hospitalization;
• lower levels of aggressive behaviour;
• more sustained play;
• and greater engagement and less negativity with parents.
For parents positive effects include:
• Greater warmth and supportiveness to children and less detachment;
• more time playing with children;
• more stimulating home environments;
• more language learning and reading support for children;
• less spanking with a wider range of discipline strategies;
• a higher likeliness to be employed or in training;
• and delayed subsequent child bearing compared to controls.
Effect sizes were modest, generally in the ten to 20 per cent range, and there were notable differences in the
effects for different groups of parents. Where parents were enrolled in EHS in pregnancy rather than later, there
613318 Deliverable 4.1
CARE Project Review of research on impact of ECEC
CARE: http://ecec-care.org/ - page 14
were stronger impacts, and early implementation had stronger effects. Generally, effects were stronger for
African-American than other ethnic groups, with only small impacts for white families. The effects of EHS were
strongest for families with a moderate number of demographic risks (three out of five) rather than low or high
risk, but there were no significant impacts upon the highest risk families who seemed impermeable to this
intervention.
There were also differential effects for the different models of intervention. Centre-based programs had the
strongest effects on child outcomes, whereas home-based programs had the strongest effects on parenting
outcomes. The mixed model combining both centre-based provision with home visiting had the most wide-
ranging and strongest positive impact.
For cognitive outcomes Barnett (2008) reported that EHS has small effects at ages two and three, (0.10 to 0.15
standard deviations - S.D.) and by age five, no significant effects remained. Aos, Lieb, Mayfield, Miller, and
Pennucci (2004) reported an adjusted effect size of EHS on test scores of 0.085 S.D., and Love et al. (2005)
estimated the impact of EHS at 0.10 to 0.13 S.D. for the cognitive and language development outcomes, similar
findings were reported by Gormley (2007) and Karoly, Kilburn, and Cannon (2005). Findings from the EHS
prekindergarten follow-up conducted by the Administration for Children and Families (2006) suggested that the
effect size of the program on receptive vocabulary for Spanish speakers was 0.27 S.D, while Karoly et al. (2005)
report that the effect size of the EHS program on cognitive outcomes near or in elementary school was 0.10 S.D..
Regarding non-cognitive outcomes, Barnett (2008) found EHS had small effects on parent outcomes. When
children were five, small improvements (0.10 S.D.) occurred in measures of children’s behavior problems,
parenting, and maternal depression. Also Barnett (2011) found that at ages two and three, some social-
emotional benefits occurred 0.10 to 0.15 S.D.. According to Love et al. (2005) EHS produced positive impacts on
several aspects of children’s social-emotional development, including less aggressive behaviour (effect size - ES=
0.11), more sustained attention (ES= 0.16), and higher engagement of their parent during play (ES= 0.20). Also
findings by the Administration for Children and Families (2006) showed that the effect size on behaviour
problems was -0.10 S.D. (i.e. less behaviour problems) and 0.12 S.D. on approaches to learning. Adams, Tout, and
Zaslow (2007) report that at 36 months, after being assigned to participate in EHS, children showed higher scores
on overall mental development. Parents rated children who have participated in EHS as showing less aggressive
behavior. During observations the children seemed to be more engaged in play with a parent and showed more
sustained attention to objects during play.
INFAN T HEALTH A ND DEV E LO P M ENT PROGRAM (IHDP )
The Infant Health and Development Program (IHDP) was an intervention aimed at improving the health and
development of premature, low birth weight (LBW) (less than 2.5 kg) infants through a combination of education
and support for parents plus enriched educational day care and health services for children. A RCT was used at
eight sites to examine the impact of IHDP on children's growth and development from birth to eight years of
age. The results of the study differed markedly by child's birth weight. For children in the range 2 to 2.5 kg, there
were large significant benefits of the enriched educational day care intervention. For the very LBW (less than
2kg) infants results were more equivocal, but with limited evidence of benefit from the intervention (Brooks-
Gunn et al., 1994; McCarton et al., 1997; Ramey et al., 1992). Also there is recent evidence that the positive
effects of the IHDP intervention are moderated by child temperament in that positive effects are most
pronounced for children rated highly for negativity in infancy (Blair, 2002).
For cognitive outcomes, Barnett (2008) reported that, at age three, the treatment group had a higher
intelligence quotient (IQ) (10 points, ES= 0.67). The IQ gain was larger (14 points, 0.93 S.D.) for the heavier LBW
program group. Although no effects persisted beyond age three for the sample as a whole, long-term gains were
found for the heavier LBW IHDP participants. The analyses indicated cognitive advantages of about 0.30 to 0.45
S.D. at ages five and eight, and higher math (but not reading) achievement at age eight. Also Magnuson and
Waldfogel (2005) reported that the heavier LBW children had IQ scores close to four points higher than their
counterparts in the comparison group at ages five and eight.
613318 Deliverable 4.1
CARE Project Review of research on impact of ECEC
page 15 CARE: http://ecec-care.org/
However, Aos et al. (2004) reported that the adjusted effect size of the program on test scores was only 0.033
S.D.. While Barnett (2008), noted that, at age 18, the heavier LBW participants scored higher for a measure of
cognitive ability and language (0.25 S.D.), as well as higher math achievement (0.34 S.D.), but not on overall IQ.
For non-cognitive outcomes, short-term positive effects on children’s behaviour were reported by Magnuson and
Waldfogel (2005), with evidence that at age three, the treatment group had fewer behaviour problems and
better maternal employment (Barnett, 2008). For long-term outcomes, at age 18, the heavier LBW program
participants had less self-reported risky behaviour (Barnett, 2008) however, no significant effects were found on
arrest rates. However, Karoly et al. (2005) reported that the effect size of the IHDP on child behaviour checklist at
age five was only -0.06 S.D..
MILWA U KEE PROJECT
The Milwaukee Project was an intervention program designed to facilitate intellectual development of very
young, disadvantaged children. The project, which began in the 1960s, aimed to improve the IQs and scholastic
achievement of children at risk, and to study the effects of intellectual stimulation on children from deprived
environments. The intervention technique employed an intensive educational program for the very young high-
risk child, beginning before six months of age. Using survey data, maternal IQ (lower than 80) was designated as
the basis for selection of a group of 40 newborns from one deprived district. Although this district contained only
three per cent of the city’s population, 33 per cent of all children who had been labeled ‘mentally retarded’ lived
there.
This very small-scale intervention included a full-time, child-oriented, centre-based program from infancy to age
six years with increasing educational input as age increased; and vocational training, childcare and household
guidance was provided for mothers. Families were randomly assigned to intervention or control groups.
Regarding the major findings, by age six all of the children from the experimental group had higher IQs than all
of the children from the control group (Garber, 1988; Heber, Garber, Harrington, Hoffman, & Falender, 1972).
Mean IQ was 120 in the experimental group and 87 in the control group. After leaving the program their IQs
started declining. By ten years of age the mean IQ of the children from the experimental group was 105, while
that of the control group was 85. At age 14, the children in the experimental group had a mean IQ ten points
above that of the control group (0.67 S.D.), but the scholastic achievement scores of the experimental group
were not better than those of the control group (Barnett, 2008). Surprisingly, both groups performed similarly in
school and as would be expected from children with a mean IQ of 80. Yet the effect on reading achievement was
an estimated 0.68 grade equivalent, or ten percentiles (Barnett, 2008).
ABECEDARIA N PROJECT
The Abecedarian Project involved a poor African-American population in North Carolina, (Campbell & Ramey,
1994; Ramey & Campbell, 1991; Ramey et al., 2000). The 111 children, whose mothers had a low IQ and low-
income, were randomised into two groups. One group was placed in a program involving centre-based care and
home visits from three months of age and continuing until children entered school. The control group received
family support, social services, low-cost or free pediatric care, and child nutritional supplements but no
additional childcare beyond what the parents and the local services provided. The high-quality ECEC program
had one qualified early childhood educator for every three infants and toddlers until age three and one for every
six children over age three.
By age 21, when 104 of the original 111 were measured, the RCT revealed that the program group, as compared
with the control group, showed gains in cognitive functioning, academic skills, educational attainment,
employment, parenthood, and social adjustment, and the earlier the start the greater the effect. The likelihood
of retention in grade during primary school declined by almost 50 per cent for children from the program
(Ramey et al., 2000). Also the mothers in the intervention group became better educated and were more likely
613318 Deliverable 4.1
CARE Project Review of research on impact of ECEC
CARE: http://ecec-care.org/ - page 16
to become employed, hence both generations benefited (Campbell, Ramey, Pungello, Sparling, & Miller-
Johnson, 2002; Clarke & Campbell, 1998).
Similar effects on cognitive outcomes were reported in other analyses of the Abecedarian data. Magnuson and
Waldfogel (2005) reported that Abecedarian children continued to outperform the comparison group on IQ tests
at age eight by just over one-third of a standard deviation. While Karoly et al. (2005) reported that the effect size
of the Abecedarian on cognitive outcomes in elementary school was 0.71 S.D..
At age 21 effect sizes were 0.23 S.D. for reading, 0.20 S.D. for math and 0.03 S.D. for high school completion
(Reynolds, Magnuson, & Ou, 2006, 2010). In Gormley’s study (2007) effect sizes were 0.53 S.D. for reading and
0.65 S.D. for math for Abecedarian children at the age of 15.
Karoly et al. (2005) reported the effect size of the Abecedarian on cognitive outcomes in elementary school was
0.71 S.D., and that for the 15 year-olds the effect size on grade retention was -0.48 S.D. and on special education
was -0.49 S.D. (i.e. both improved), while the effect on high school completion was surprisingly low at 0.06 S.D..
In another report, Campbell et al. (2002 cited in Barnett & Masse, 2007) found that at age 21 there were
continued effects on IQ and achievement, though effects on IQ appear to have declined and then stabilized at
about five points compared to early assessments. Effects on school success include much lower levels of grade
retention; placements in special education classes; reduced high school dropout and a higher rate of attending a
four-year college at age two. The authors also report that there are differences between treatment and control
groups in terms of reading achievement at age 14/15; math achievement at age 14/15; grade repetition; special
education; and college attendance. Similarly Barnett (2008) estimated that initial gains in IQ have declined over
time, with effect sizes of about 0.33 S.D. at ages 15 and 21. Effects on reading and math achievement averaged
about 0.40 S.D. from ages eight to 21, with only a slight decrease over time. Similar effects are reported by
Temple and Reynolds (2007), and Pianta et al. (2009) in that this educationally focused childcare intervention
reduced grade retention and placement in special education by 23 percentage points each. Also high school
graduation and attendance at college were significantly affected. In addition, according to Adams et al. (2007)
children in the intervention group had higher achievement in reading, math, and cognitive assessment scores
through young adulthood. As young adults, participants in the intervention had completed more years of
education, and were more likely to have participated in college. So overall there is a consistent positive message
on the long-term impact of Abecedarian on cognitive and educational outcomes.
In addition to cognitive effects the Abecedarian Project produced long-term effects for non-cognitive outcomes.
Manning, Homel, and Smith (2010) in a meta-analysis showed effects of 0.49 S.D. (on deviance), 0.43 S.D. (social
participation) and 0.21 S.D. (criminal justice). McLaughlin, Campbell, Pungello, and Skinner (2007) showed that
the early intervention reduced depression, and Barnett (2008) reported that the program group mentioned
fewer depressive symptoms at age 21 (0.42 S.D.). Also according to Karoly et al. (2005) the effect on adult crime
and delinquency by age 21 was 0.13 S.D. whereas effects for employment were 0.28 S.D.; 0.53 S.D. for skilled
jobs and 0.25 S.D. for use of social services.
Barnett (2008, 2011) reports positive effects were found for health-related behaviors and symptoms of
depression. The program group was more likely to have a skilled job, less likely to have become teen parents, and
less likely to smoke marijuana. Effects were not found on social development or behavior during the program or
in later delinquency and crime, as was also reported by Temple and Reynolds (2007). However, control group
involvement in crime and delinquency was low. Finally, the free childcare appeared to have improved mothers’
long-term employment opportunities and earnings. Similarly, Pianta et al. (2009) comment on the long-term
effects beyond schooling and cognitive skills in a similar way.
PROJECT C ARE
The same team involved in the Abecedarian Project undertook a subsequent RCT study (Project CARE) that
compared the effects of a centre-based program, home visiting and control condition with interventions starting
shortly after birth, again with low-income African-American families. At 12, 18, 24, and 36 months, the day care
plus home visit intervention group scored significantly higher on developmental assessments than the control
and home visit only groups. At 30, 42, 48, and 54 months, the two intervention groups differed from each other
613318 Deliverable 4.1
CARE Project Review of research on impact of ECEC
page 17 CARE: http://ecec-care.org/
in that the home visit only group's scores were lower than the day care plus home visit group's scores. Children
in treatment groups that included childcare were rated as more task-oriented in infancy and tended to show
higher, more stable cognitive scores beginning during late infancy and continuing through early childhood than
the children who did not receive the childcare intervention (Ramey & Campbell, 1982; Wasik, Ramey, Bryant, &
Sparling, 1990). In essence, only the centre-based program had any significant effect (Burchinal, Campbell,
Bryant, Wasik, & Ramey, 1997; Wasik et al., 1990).
ECEC FOR C HILD REN AG ED O VER THRE E Y E ARS
There are also a range of programs involving ECEC for children from three years upwards that have been used as
an intervention for children from disadvantaged backgrounds. The evaluations of these programs include RCTs
and quasi-experimental studies.
RANDOMIS E D CONTROL T RIALS (RCTS)
PERRY PRESCHOOL PROJECT (PPP)
The Perry Preschool Project (PPP) occurred in Ypsilanti, Michigan. This half-day, five days a week, centre-based
program started at three years of age and was supplemented by 90-minute weekly home visits. It was based in
an area of extreme urban deprivation and the population was African-American. Children with IQs lower than 90
were randomly assigned to the intervention or control groups, and 123 of the children have been followed into
adulthood. The intervention involved a high-quality educationally oriented curriculum (High/Scope), with well-
trained staff. In a RCT the program was demonstrated to have long-term effects. In school, the intervention
group showed higher levels of educational achievement, but there were no long-term effects for IQ. By age 27,
the long-term benefits of the intervention included: reducing school drop-out, reducing drug use, reducing
teenage pregnancy, enhancing employment, reducing welfare-dependence and reducing crime. Fewer females
in the intervention group showed ‘educable mental impairment’ or poor mental health, and the males had far
fewer criminal arrests (Schweinhart, Mcnair, Barnes, Larner, & Mary, 1993).
Barnett (2008) reports that initial effects of PPP on language and general cognitive abilities after two years were
about 0.90 S.D., which is about the size of the typical black/white test score gap. The standardized effect size for
achievement gains from pre-k at age four in cognitive/language development is 0.75 S.D. (Barnett, 2010).
According to Barnett (2011) treatment and control groups did not differ on measured IQ at the start, but at the
end of the program, the preschool group scored 0.87 S.D. higher than controls, but the IQ gain disappeared by
age eight.
Long-term effect sizes were in the range of 0.30 to 0.50 S.D. and high school graduation increased from half to
two-thirds (Barnett, 2008). There were positive effects on achievement tests (e.g. 0.33 S.D. on reading and math
at age 14 through age 27. Barnett (2011), and Nores, Belfield, Barnett, and Schweinhart (2005) reported that
participation in preschool was correlated with associate or higher degree by age 40. Finally, Karoly et al. (2005)
reported that effect of the PPP on grade retention was -0.15 S.D. by age 27 and on special education was -0.29
S.D. by age 19, the effect on high school completion was 0.37 S.D. by age 40.
Belfield, Nores, Barnett, and Schweinhart (2006) maintain that program participation is linked with lower grade
retention and less placement in special education classes. In terms of further and higher education program
males reported fewer semester credits, but program females reported higher rates of college progression. Also
authors indicate that individuals continued to accumulate education credentials after age 28. Similarly, both
Temple and Reynolds (2007) and Pianta et al. (2009) report better outcomes for the participants in the PPP in
terms of high-school completion, reduced special education, highest grade completed, and college attendance.
For non-cognitive outcomes, Barnett (2008) reported that the preschool group had better behaviour as reported
by teachers and less involvement in delinquency and crime. Long-term effect sizes are in the range from 0.30 to
613318 Deliverable 4.1
CARE Project Review of research on impact of ECEC
CARE: http://ecec-care.org/ - page 18
0.50 S.D.. The number of arrests by age 27 fell by half, and employment at age 40 showed an increase of 14
percentage points. While Karoly et al. (2005) reported the effect size on number of arrests by age 27 was -0.54
S.D., on per cent employed was 0.25 S.D. at age 27 and 0.30 S.D. at age 40, and on monthly earnings ($1,993) at
age 27 was 0.51 S.D.. In terms of social services use the effect size was 0.44 S.D. on per cent received in the past
ten years by age 27 and -0.37 on per cent of lifetime use of social services by age 40. Similarly, Manning et al.
(2010) reported effect sizes on deviance of 0.64 S.D. and on criminal justice of 0.41 S.D.. While Muennig,
Schweinhart, Montie, and Neidell (2009) used 37 years of follow-up data to explore the linkage with adult health.
The intervention led to improvements in education, health insurance, income, and family environment, which, in
turn, led to improvements in behavioural risk factors and health.
Belfield et al. (2006) reported that at age 40, the program group was more likely to be employed, have higher
earnings, and rely less on economic support from family or friends. In terms of lifetime gross earnings, Belfield et
al. (2006) reported that for program males, lifetime earnings ranged from $874,608 to $1,391,307; these totals
compare favorably to control group males’ earnings of $651,296 to $1,168,871. The program differential is
considerable, with earnings 11 per cent to 34 per cent higher than the control group. Similarly, program females
reported higher lifetime earnings; the absolute earnings differential may even be higher than the males, and the
premium ranges from 19 per cent to 36 per cent. Moreover, Belfield et al. (2006) reported higher rates of asset
possession and money amounts of wealth by the program group, which has implications for wealth and well-
being on retirement. High/Scope Perry Preschool Program was also found to have affects as 48 per cent of the
no-program group had never been arrested for violent crime, compared to 32 per cent for the program group;
similarly 17 per cent of the no-program group were incarcerated at the time of interview, compared to six per
cent of the program group. Also, there were program differences in murder rates (2% vs. 5%). Overall, there was
lower lifetime criminal activity by the program group.
Turning to welfare receipt, Belfield et al. (2006) reported that in general, welfare reliance was slightly lower for
program participants. This differential was large for the ages 17 to 27, but narrowed subsequently with
equivalent proportions reporting any services during the ages 33 to 40. Also, welfare receipt differed
fundamentally by gender, particularly when females were the primary caregivers for children. There were also
health status differences across the groups, with the program group being less likely to report that they: had
stopped working for health reasons (43% vs. 55%); had a health problem (20% vs. 29%); smoked (42% vs. 55%);
used soft drugs (45% vs. 54%); used hard drugs (22% vs. 29%); or needed treatment for drug-use or drinking
(22% vs. 34%). Also, there was a difference between the groups in terms of mortality rates. Of the initial 58
program participants, one female and one male were deceased by age 40; of the 65 participants in the no-
program group, two females and three males were deceased. Finally, the authors suggest that the data showed
differences in family formation and behaviors with intergenerational consequences, such as abortions (17% vs.
32%) and differences in family size and two-parent family rates and teenage parenting.
Finally, Pianta et al. (2009) suggest that through age 40, the program was associated with increased employment
and earnings, decreased welfare dependency, and reduced arrests. The number of arrests by age 27 fell by half,
and employment at age 40 increased by 14 per cent. Authors report the following percentages: ever arrested as
juvenile 16% vs. 25%; mean number of adult arrests 2.3 vs. 4.6 (age 27); and adult smoker 42% vs. 55% (age 40).
EARLY TRAINING PROJECT (ETP)
The ETP occurred in Murfreesboro, Tennessee (1962 to 1964). Sixty-five, three to four year-old children were
randomly assigned to treatment (44) and control (21) groups. Children were selected if they lived in poor or
deteriorating housing or public housing, had a low family income, and had parents with less than a high school
education and in an unskilled or semi-skilled occupation. The intervention program consisted of a ten-week
summer preschool program for the two or three summers prior to the first grade, plus weekly home visits during
the remainder of the year. Hence this intervention was comparatively of low duration. The purpose of the study
was to assess the impact of a preschool educational intervention on attitudes relating to achievement, and on
academic performance. Children received positive reinforcement and participated in activities focusing on
motivation and persistence in classes of four to five. They also received a 90 minute home visit per week
(Anderson, 2005).
613318 Deliverable 4.1
CARE Project Review of research on impact of ECEC
page 19 CARE: http://ecec-care.org/
ETP data comes from three sources: interviews with subjects and parents, program administered tests, and
school records. Data from this study include tests of intellectual development prior to, during, and after
intervention; tests of school achievement from first grade to high school; various indices of the affective domain;
school records; ratings by teachers and counselors; interviews with participants in 1976 and 1979; annual
interviews with the parents from 1962 to 1966 and again in 1975; and demographic and family data. No crime
data were collected (Anderson, 2008).
According to Karoly et al. (2005) the effect of ETP by age 18 on grade retention was 0.12 S.D. and on special
education was -0.79 S.D. and at age 18, 0.27 S.D. on high school completion, thus indicating overall beneficial
effects from the ETP.
HEAD START
Head Start is different from other interventions. It is federally funded but administered by each state
independently. As a consequence Head Start varies substantially between states, and also within states. Head
Start is a broad-based early intervention program to improve outcomes for children in disadvantaged families. It
was initiated in the 1960s as an eight-week summer preschool program but rapidly developed to be a year-
round program and has included a wide range of variations. Typically a Head Start program would include
centre-based early childcare and education from three years of age on at least a half-time basis. A range of other
services may supplement this basic package and the diversity has made it difficult to assess. By the early 1970's
Head Start had become a continuous preschool program, and by 1999 it served 800,000 children at a cost of
$5,400 per child.
Participation in Head Start has been associated with short-term improvements in cognitive development (see
Barnett, 1995; Karoly et al., 1998 for reviews). However often effects appeared to fade after a few years.
However, subsequent follow-up in adolescence indicated that Head Start was still having an effect, possibly
'sleeper' effects, in that Head Start graduates showed higher educational attainment. Oden, Schweinhart,
Weikart, Markus, and Xie (1996) conducted a 17-year follow-up study. Once background differences were
adjusted, Head Start subjects were generally equal to or better in educational development than what they
would have been without Head Start. Kresh (1997) synthesized 30 years of research on the effects of Head Start.
Findings indicated that Head Start had a substantial, immediate effect on participants, but the long-term effects
were less evident. There was some evidence that Head Start increased parent-child communication, parental
participation in school, mothers' satisfaction with their quality of life, and confidence in their coping abilities.
Head Start participation decreased maternal depression, anxiety, and somatic symptoms. Head Start was
associated with some community effects including increased educational emphasis on the poor and needy,
greater sensitivity in health service delivery, and increased employment.
The Head Start initiative was a source of many studies on the effects of preschool education. Some were RCTs
and some were quasi-experimental evaluations. McKey et al. (1985) developed a meta-analysis of 210 studies
evaluating Head Start programs. They concluded that Head Start programs have an immediate positive effect on
child development, but these effects 'wash out' after two years. However many studies were poorly controlled.
In one of the more robust studies, Lee, Brooks-Gunn, and Schnur (1988) considered data on 969 children and
nineteen preschools. Some children had attended a Head Start preschool, some had attended other preschools
and some children had not attended preschool. They found evidence of beneficial preschool effects upon
cognitive measures, with the greatest effects occurring for the most disadvantaged children.
However such summaries have not been aware of ethnic variation in Head Start effects. Currie and Thomas
(1993, 1995) used the National Longitudinal Survey of Youth (NLSY, a nationally representative US cohort) data
to evaluate Head Start. They compared children who attended Head Start with siblings who did not. This
strategy provided a means of controlling for family and other background factors. Using this nationally
representative sample they found substantial gains in literacy, numeracy and grade repetition for white and
Hispanic children, but not African-American children, at eight years of age, associated with Head Start. For
African-American children these gains faded out over the early school years. Head Start also appeared to
positively influence the immunisation rates, growth and nutritional status for African-American children, with
those children attending Head Start being taller than their siblings who did not. For white children the
613318 Deliverable 4.1
CARE Project Review of research on impact of ECEC
CARE: http://ecec-care.org/ - page 20
educational gains persisted into adolescence. This suggests that the fade out was associated with African-
American children's experiences in school. This explanation is supported by evidence from Currie and Thomas
(1998) that African-American children attending Head Start go on to lower quality schools than other African-
American children. This is not true for white children.
Lee and Loeb (1995) in a follow-up study of Head Start participants provide a possible explanation for fading
effects i.e. that children who attend Head Start are more likely to attend elementary schools of lower
educational quality, with a less favourable socio-economic composition of students, and more problems of
safety, which may cancel out previously established positive effects.
Garces, Thomas, and Currie (2000) used NLSY data to consider the effects of Head Start for young adults. They
found that Head Start had positive effects on educational outcomes and earnings for whites but not African-
Americans. White graduates of Head Start showed an increased likelihood of graduating from high school, and to
have higher earnings. For African-Americans attendance at Head Start was significantly associated with lower
criminal activity. This was not so for whites. These results indicate that interventions such as Head Start have
varying effects dependent upon the population and context involved. Other evidence supports the view that
versions of the program involving parents improved children's outcomes (Lee, Brooks‐Gunn, Schnur, & Liaw,
1990).
A United States General Accounting Office (1997) report concluded that, after the first 30 years of Head Start,
very little was known about the impact of the program. Only 22 out of 200 studies utilised any comparison
group. These studies indicated higher gains in self-help, academic skills and cognitive development in the short-
term, but there was inconsistent support for longer-term effects. There was some evidence of health-related
benefits with Head Start participants being more likely to receive preventive health services. The dearth of strong
evidence led to the setting up of two systematic evaluations. The Family and Child Experiences Survey (FACES)
project is following a random sample of 3,200 families from 40 representative Head Start programs. This study
(Zill et al., 2001) reports significant positive effects for Head Start on vocabulary, literacy, numeracy and social
skills at the start of school, with effects being greater for the most disadvantaged children (McKey, 2003). They
also report that independent observers rate the quality of Head Start programs as generally high, with some
indication of better child outcomes being associated with higher quality programs. Also there are some benefits
for parents in terms of increased employment and decreased benefit dependence. In a review Burger (2010)
presents the results of the Head Start Family and Child Experiences Survey (FACES), at age five to six years as
showing significant positive effects on vocabulary, early math and writing, with effects ranging from 0.05 to 0.67
S.D..
Barnett (2010) reports effect sizes for achievement gains from pre-K: cognitive/language 0.09 S.D. at age four and
0.18 S.D. at age three; with math 0.15 S.D. at age three; and print 0.25 S.D. at age four and 0.24 S.D. at age three.
The Head Start Impact Study reported by Gormley (2007) shows effect sizes for pre-reading of 0.24 S.D. for three
year-olds, and 0.22 S.D. for four year-olds; and the effect size for spelling was 0.16 S.D. for three year-olds with
no impact for four year-olds. According to Barnett (2011), after one year of Head Start at age three or four, 13 of
22 measures of language, literacy, and math effects were significant; averaging 0.18 S.D. In addition, Barnett
(2008) reported that the estimated cognitive effects of nine months of Head Start ranged from 0.05 to 0.25 S.D..
A smaller RCT by Abbott-Shim et al. (2003 cited in Barnett, 2008) found gains of 0.32 S.D. on vocabulary
(compared to Head Start Impact Study: 0.05 to 0.12 S.D.). Barnett (2008) concluded that one year of Head Start
has initial effects on cognitive abilities that are in the range of 0.10 to 0.30 S.D. While Ludwig and Phillips (2008)
estimate effects of between 0.04 to 0.43 S.D. on a range of cognitive, linguistic and educational outcomes at age
four years.
A RCT by the US Department of Health and Human Services estimated impacts of around 0.1–0.2 S.D. one year
after Head Start entry (Puma et al., 2005). The study revealed short-term test score impacts as large as 0.20 S.D..
RCTs of longer-term impacts of the program have been estimated at 0.28 S.D. for the most disadvantaged
children (Deming, 2009). According to Reynolds et al. (2006, 2010) the effect of the Head Start Follow Through
Program at ages 12 to 15 ranged from 0.00 to 0.17 S.D. for reading and from 0.13 to 0.26 S.D. for math.
The Head Start Impact study was a nationally representative multisite RCT study to assess the impacts of Head
Start on children and families through to the third grade (age nine years). Nearly 5000 newly entering children
613318 Deliverable 4.1
CARE Project Review of research on impact of ECEC
page 21 CARE: http://ecec-care.org/
(age three or four) from poor families were randomly assigned to the intervention group with access to the Head
Start program services, or the control group without access to Head Start but could receive other early childhood
education services selected by parents rather than only receive parental care (U.S. Department of Health and
Human Services & Administration for Children and Families, 2010). After attending one year of Head Start, age
three cohort children benefitted in all the four domains examined (cognitive development, social/emotional
development, physical development and parenting practices); for the age four cohort children, positive effects
were found in language and literacy elements of the cognitive domain and access to dental services in the health
domain. However, at the end of first grade, only a few significant differences in outcomes remained, including a
favourable impact for the four year-old cohort on reading, but an unfavourable impact for the three year-old
cohort on grade promotion. The fact that many of the control group received ECEC rather negates the point of
the study and makes conclusions difficult.
It is clear, however, that the effects of Head Start programs show great variability. Bloom and Weiland (2015)
used data from the Head Start Impact Study to analyse variation in Head Start effects across children, subgroups
of children, and Head Start sites. They found that:
Head Start produced a ‘compensatory’ pattern of effects that increase cognitive outcomes most for
children with the weakest initial cognitive skills. This tended to equalize cognitive skills across
program participants.
Head Start increased cognitive outcomes far more for dual language learners and Spanish-speaking
children than for other children.
Much of the program’s effect represents ‘compensation’ for limited prior English.
The ‘value added’ by any Head Start program depends on both the program itself and the quantity and
quality of other local options for early child education.
Some Head Start centers are much more effective than their alternatives, while others are much less
effective, with most centers operating between these extremes.
A policy brief from the National Forum on Early Childhood Programs and Policies (2010) summarises much
research on Head Start. Evidence suggests that the achievement of children who applied but were not randomly
assigned to a spot in a Head Start classroom had caught up to Head Start students’ achievement levels by first
grade. However The ECEC experiences of treatment and control groups were not distinctly different, in that
many of the control group had enrolled in ECEC and the more similar the ECEC experiences of the control and
treatment groups, the less likely it is that the two groups of children will differ in terms of their outcomes. Also
dual-language learners and children with special needs benefited more from Head Start participation than other
groups, and the benefits for these groups persisted beyond first grade. There is room for improvement in Head
Start. Less than one in 20 children were in centres with an excellent quality rating and only about half were in
centres with recommended pupil-staff ratios.
Overall the very mixed results of the RCT evaluations of Head Start reflect the very mixed nature of the program
itself in its myriad of locations around the USA. In particular, the fact that in several studies the comparison
(control) group also received unspecified ECEC makes findings of null effects questionable. The mixed results
probably reflect differences in study design, lack of control of what happens to the comparison (control) group,
different measurements and methods as well as differences in implementation affecting quality.
613318 Deliverable 4.1
CARE Project Review of research on impact of ECEC
CARE: http://ecec-care.org/ - page 22
QUASI- EXP E RIM E NTA L EVALUAT ION S OF HEAD STA RT
COGNITIVE OUTCOMES
According to the Head Start Impact Study (U.S. Department of Health and Human Services & Administration for
Children and Families, 2010), the effects for the four year-old entry cohort at end of first grade were only for
vocabulary (0.09 S.D). For the three year-old entry cohort at first grade it was found that the program had an
effect only on oral comprehension (0.08 S.D.). However, according to Barnett (2008) another study of Head
Start’s initial effects in Tulsa, Oklahoma revealed that for one year of Head Start at age four, effects were 0.33 to
0.55 S.D. on literacy and math assessments. Inconsistency in results for Head Start continues to prevail.
Domitrovich et al. (2013) examined the exposure to an enhanced Head Start model and academic functioning in
disadvantaged children at kindergarten. The mean effect size across language and literacy measures comparing
children with one versus two years was 0.36 S.D and for numeracy skills the effect size was significantly higher
(0.33 S.D.).
Regarding studies comparing siblings who attended, or not, Head Start, Magnuson and Waldfogel (2005)
reported that six year-old Head Start children scored close to seven percentile points higher on a vocabulary test
than their siblings who did not attend preschool.
NON-COGNI T IVE OUTCOMES
Barnett (2008) reported that after nine months of Head Start attendance, behaviour problems and hyperactivity
were all significantly reduced (0.13 to 0.18 S.D.) for three year-olds. Access to dental care was improved and child
health, as reported by parents, was modestly improved (0.12 S.D.) for three year-olds.
Gormley (2007) reported that Head Start reduced the incidence of problem behaviour among three year-olds,
with an overall effect size of 0.18 S.D.. Turning to health care impacts, Head Start participation had positive
impacts on the child’s health status, especially for three year-olds. For overall health status, the effect size was
0.12 S.D. for three year-olds, with no impact for four year-olds. Head Start participation had stronger positive
impacts on the child’s dental care. The effect size was 0.34 S.D. for three year-olds, 0.32 S.D. for four year-olds.
QUASI- EXP E RIM E NTA L STU DIES OF OTHE R U S PR O G RAM S
CHILD- PA RENT C E NTERS (C P C)
Child-Parent Centers (CPC) was initiated in 1967 to provide centre-based educational support and family support
to disadvantaged children and their parents, including education, family, and health services and half-day
preschool and school-age services up to nine years. The guiding principle was that by providing a child-centred,
613318 Deliverable 4.1
CARE Project Review of research on impact of ECEC
page 23 CARE: http://ecec-care.org/
individualised approach to social and cognitive development in a school-based, stable learning environment
during preschool, supported by energetic parental involvement, scholastic success will follow. Reynolds and
colleagues (Reynolds, Temple, Robertson, & Mann, 2000; Reynolds, Temple, Robertson, & Mann, 2001a, 2001b)
run the federally-funded Chicago Longitudinal Study of the effects of this early childhood intervention. This
quasi-experimental study began in 1986 with a non-randomised, matched-group cohort of 1,539 (989
intervention, 550 control) low-income, mostly black children born in 1980 and enrolled in public early childhood
programs in 25 sites in Chicago. The intervention group received CPC services while the control group typically
did not receive any educational services until age five (kindergarten).
Overall, the CPC program has shown effectiveness on a range of child and adolescent outcomes, with the largest
benefits found for participation in the preschool program and fewer (but still significant) benefits found for
school-age participation. The effects were evident from grade three through grade six and nine (Reynolds, 1995,
2000; Reynolds, Temple, Robertson, & Mann, 2002). More specifically:
Children with any CPC participation (in preschool or school) outperformed children who had no CPC
exposure. Children who participated in the preschool intervention for one or two years had a higher rate of
high school completion, more years of completed education, and lower rates of juvenile arrest, violent
arrests, and school dropout.
Children with two years of preschool experience had higher cognitive readiness at age five, and in reading
and math achievement in school than those with one year of preschool.
Both preschool and school age participation were significantly associated with lower rates of grade retention
and special education.
Differences in special education placement between CPC and non-CPC groups emerged at third grade and
were significant from grade three through grade nine.
The effects of preschool were greater for boys, especially for reducing school dropout.
At age 21 the CPC preschool group had significantly better outcomes (Reynolds & Robertson, 2003; Reynolds et
al., 2001a, 2002) in terms of:
Less years of special education from ages six to 18.
Lower rates of child maltreatment from ages four to 17 and criminality.
High school completion.
Highest grade completed.
Finally, these positive effects of early childhood intervention on educational attainment, social development and
criminality persisted up to age 24 (Reynolds et al., 2007). The findings confirmed the following benefits of the
program; higher school completion, grade completion, longer college attendance; lower criminality, lower
depressive symptoms.
Short-term effects include moderate to high effect sizes in a number of studies. Barnett (2008) reported that
estimated effects on test scores at kindergarten varied from 0.35 to 0.77 S.D.. Similarly, Magnuson and Waldfogel
(2005) reported that children who attended CPC during the year before kindergarten scored 0.64 S.D. higher on
academic skills in kindergarten. Karoly et al. (2005) reported the effect size of CPC on cognitive outcomes near or
in elementary school was 0.35 S.D.. Pianta et al. (2009) reported that estimated effects on test scores at
kindergarten entry, with effects of just one year of CPC attendance equal to between 25% and 85% of the
achievement gap at school entry. Finally, Gormley (2007) maintains that effect sizes for test scores at school entry
ranged from 0.46 to 0.63 S.D..
For longer-term effects upon cognitive and educational outcomes, Karoly et al. (2005) compared children in
grades three, five, and eight who had attended CPC with children who attended ordinary preschool in reading
and math. CPC children outscored the comparison group every year, with effect sizes from 0.17 to 0.34 S.D.. In
addition, the effect on grade retention was -0.34 S.D.by age 15, on special education -0.26 S.D. by age 18, and on
school completion by age 20 was 0.23 S.D.. (Similar results are reported by Barnett, 2008; Gormley, 2007;
Manning et al., 2010; Pianta et al., 2009; Temple & Reynolds, 2007).
According to Barnett (2008) the CPC study found positive effects on the following outcomes: cognitive test
scores through at least middle school, special education, reduction in grade retention and more high school
613318 Deliverable 4.1
CARE Project Review of research on impact of ECEC
CARE: http://ecec-care.org/ - page 24
graduation. Effects on cognitive abilities decline over time, but as late as eighth grade they are still nearly 0.20
S.D.. The effects on schooling outcomes are substantial (effect sizes of 0.23 to 0.34 S.D.) with a 15 percentage
point reduction in grade retention, a ten percentage point reduction in special education placements, and an 11
percentage point increase in high school graduation by age 20. Also, Reynolds, Temple, Ou, Arteaga, and White
(2011) examining effect on age 28 found that relative to the comparison group, the preschool group had
significantly higher levels of educational attainment for four of six measures, including highest grade completed
and college attendance.
For longer-term non-cognitive outcomes, Manning et al. (2010) report that effects for socio-emotional
development effects ranged from 0.11 to 0.12 S.D, while the effect on criminal justice was 0.20 S.D. and on
family well-being was 0.11 S.D.. Temple and Reynolds (2007) report that CPC preschool participation reduced
arrests by 23.8 per cent. Finally, Karoly et al. (2005) reported that effect size of CPC on the life skills measure at
eighth grade was 0.28 S.D., and effects for crime by age 18 was -0.20 S.D. and for violent offenses -0.19 S.D..
Reynolds et al. (2011) summarise the CPC findings succinctly. There were consistent and enduring benefits for
children who began preschool at age three or four (compared with children who began kindergarten when
older), and especially for males and children of high-school dropouts. In particular, by age 28, the former
preschool students had higher educational levels, incomes, socioeconomic status, and rates of health insurance
coverage—and lower rates of substance abuse and legal problems—than the kindergarten students.
MIAMI S C HO OL READIN E SS PROG R A M
The Miami School Readiness Program is a large-scale, university–community project. It primarily serves Latino
children to prepare children for school, particularly those from low-income families. Priority for school readiness
program participation is given to: children whose parents are economically disadvantaged, children at risk of
abuse, neglect, or exploitation, and children with special needs. Services vary based on individual need.
Ansari and Winsler (2012) report that children who attended centre-based ECEC improved more compared with
children who attended individual family-based ECEC, who did not improve much in language skills. Females
improved more than males. Also children’s social skills improved over time, with teachers rating children as
having better social skills in family-based ECEC than children in centre-based ECEC. However teachers also rated
children in family-based ECEC with more behavioural concerns than those in centre-based ECEC.
GREAT STA R T READINES S PROGRAM
Following the success of Head Start and other projects demonstrating positive effects for preschool attendance,
many states set up their own preschool programs. State pre-kindergarten programs (also called state pre-K)
provide state-funded, classroom-based educational services to young children, typically four year-old children,
although some states also enroll three year-old children. Some programs are for low-income children or others at
risk while some are universally open to all children. Universal pre-K programs are considered later. Here we focus
on programs targeted on disadvantaged children and the evaluation involves quasi-experimental studies.
Michigan started a preschool program for children at risk of school failure in 1985. An evaluation comparing
program children with non-program children from similar backgrounds found evidence of positive effects.
Teacher ratings indicated improved interest in school, and attainment on a wide range of subjects. Program
children were also 35 per cent less likely to be retained in grade (Michigan Department of Education, 2002).
The Great Start Readiness Program, formerly the Michigan School Readiness Program (MSRP), is a state-funded
preschool initiative which began as a pilot program in 1985. To qualify for the program, a child must be four years
of age and have at least two of 25 risk factors, for example, living in a low-income family and/or in a single parent
family (Xiang & Schweinhart, 2002). Children in the MSRP receive a child developmental preschool program that
provides age-appropriate activities in order to promote their intellectual and social growth and school readiness.
613318 Deliverable 4.1
CARE Project Review of research on impact of ECEC
page 25 CARE: http://ecec-care.org/
Children's families receive parenting support, guidance, and referrals to community services as needed (Xiang,
Schweinhart, Hohmann, Smith, & Smith, 2000).
The first year of the longitudinal evaluation uncovered some promising findings (Florian, Schweinhart, & Epstein,
1997). Intervention children exhibited behaviours more conducive to learning in kindergarten (e.g. completing
assignments and retaining learning) than similar children without a preschool experience, according to teachers’
ratings. Children that attended the program were more advanced in six areas of child development: initiative,
social relations, creative representation, music and movement, language and literacy, and logic and mathematics,
compared to the control group, and intervention children were rated as more advanced on 19 of 30 items.
Findings from the third year evaluation indicated that after controlling for site differences, participants'
characteristics, and socio-economic status, intervention children remained significantly higher in overall
development in kindergarten and on some aspects of school readiness, compared to non-intervention students
(Xiang et al., 2000). Intervention children had significantly lower grade retention than comparison children.
For the fifth year findings Xiang and Schweinhart (2002) suggest that in grade four, students who had attended
the program had a significantly higher percentage of satisfactory scores on academic performance than students
who had not attended the program. Larger percentages of the intervention group demonstrated proficiency in
both math (55% vs. 47%) and reading (43% vs. 35%). Again, a smaller percentage of the intervention group than
comparison group had ever repeated a grade (14% vs. 22%). Parents of intervention students were significantly
more involved in school activities and communication with teachers during the first three years of school than
comparable parents whose children did not participate. Parent involvement, as previously, was positively
correlated to children’s social relations across years, and with fourth grade academic performance.
Lamy, Barnett, and Jung (2005) examined the effects at kindergarten entry. The findings provide strong evidence
of the positive impact of the program on children’s language, literacy and math skills development. The effect
sizes of the impact of state-funded preschool programs on receptive vocabulary, print awareness and math
scores were 0.21, 0.96 and 0.44 S.D. correspondingly. Following children to eighth grade, Malofeeva, Daniel-
Echols, and Xiang (2007) reported evidence of a relation between program attendance and lower grade retention
rates and academic performance.
Schweinhart, Xiang, Daniel-Echols, Browning, and Wakabayashi (2012) looked at the program effects on high
school graduation, grade retention, multiple grade retention, and Michigan Merit Examination proficiency. Their
study identified better on-time school graduation – 57 per cent versus 43 per cent; lower retention in grade – 37
per cent compared to 49 per cent. Furthermore 43.5 per cent of the cost of the Great Start Readiness Program
was recouped from savings due to the reduction in grade retentions. By grade 12, fewer intervention children of
color were retained for two or more grades than their non-intervention counterparts – 14 per cent versus 28 per
cent. At grades 11 or 12, the intervention group had a higher performance in mathematics (27% vs. 22%) and in
math and language arts combined (35% vs. 28%).
TEXAS TA R G ETED PR E -KINDE R G ART EN PROGRAM
Texas began offering pre-K during the 1985-1986 academic year. The purpose of state-sponsored pre-K in Texas is
to bolster the academic performance of at risk children. The risk factors include: free or reduced-price lunch
eligibility, limited English proficiency, homelessness or unstable housing, foster care participation, or parents who
are on active military duty or who have been injured or killed on duty. The Texas program ranks low in quality in
terms of class size, staff to pupil ratios, and spending per capita (Barnett, Carolan, Fitzgerald, & Squires, 2011). As
such, an evaluation of this program's impact on student outcomes can provide guidance on whether modest
programs, perhaps all that can be hoped for in the current budgetary environment, are worth implementing.
A longitudinal study of the effects of Texas's targeted pre-K on educational outcomes (Andrews, Jargowsky, &
Kuhne, 2012). For the third grade reading test there were statistically significant effects for public pre-K
attendance for those students with economic disadvantage only, for those with limited English proficiency, while
the largest effect size was experienced by students eligible for the program due to both economic disadvantage
and limited English proficiency. Attendance in public pre-K, relative to the alternatives, significantly reduced the
613318 Deliverable 4.1
CARE Project Review of research on impact of ECEC
CARE: http://ecec-care.org/ - page 26
probability of retention in grade. The odds of retention were 24 per cent lower for those who attended public
pre-K. Similarly the likelihood of being assigned to special education in third grade was 13 per cent lower for pre-
K children.
SYRACUS E FAMILY DEVELOP M E NT RESEARCH P R O GRAM
The Syracuse Family Development Research Program was a comprehensive childcare, education, health and
family support program from pregnancy to the start of school designed to improve child and family functioning
through home visitation, parent training, and individualised day care (Lally, Mangione, & Honig, 1988). The
program operated in a single site in Syracuse, New York, between 1969 and 1976, and has undergone long-term
assessments of effects on participants. The program targeted young, African-American, single-parent, low-
income families in order to improve children’s cognitive and emotional functioning, create a positive outlook
among children, and decrease juvenile delinquency. Mothers were in the last trimester of first or second
pregnancy, at the start of the intervention. Child Development Trainers visited each family weekly and focused
on increasing family interaction, cohesiveness, and nurturing. In the Children’s Center (for day care), infants were
assigned to a caregiver for attention, cognitive and social games, sensorimotor activities, and language
stimulation. The preschool program supported child-chosen opportunities for learning and peer interaction, and
areas of the center were designated for specific types of activities. The evaluation (Lally et al., 1988) involved 190
families and found that the intervention produced better educational attainment and school attendance for girls,
but not boys, as compared with a control group. In adolescence, there were improvements in social adjustment
and reduced criminality for the intervention group.
DEL AWA R E EARLY C HILDH OO D LO N GITUDIN A L STUDY
In the mid-1990s, Delaware began a comprehensive early childhood program for all children aged four living in
poverty. The Early Childhood Assistance Programs (ECAP) were modeled on Head Start and used the Head Start
Performance Standards as their program standards. In combination with federal Head Start funding, the ECAPs
made Delaware one of the first states to provide comprehensive four year-old early childhood programming for
every child living in poverty (Gamel-McCormick, Amsden, & Hartranft, 2005).
The Delaware Early Childhood Longitudinal Study was designed as a retrospective, two-group, post-test only
evaluation design, which is a weak evaluation design, and was very small-scale. The 26 students who received
ECAP or Head Start services at four years of age were compared with 103 students not receiving ECAP or Head
Start services. Students who received ECAP or Head Start services at age four were better, in the fifth grade (age
11), at reading, math, and writing than those who did not receive ECAP or Head Start services.
613318 Deliverable 4.1
CARE Project Review of research on impact of ECEC
page 27 CARE: http://ecec-care.org/
E U R O P E A N S T U D I E S O F E C E C A S A N I N T E R V E N T I O N
F O R C H I L D R E N F R O M D I S A D V A N T A G E D B A C K G R O U N D S
There are several European countries where evaluations of such intervention have occurred and these include
the UK, Denmark, Germany, Switzerland, France and the Netherlands. Studies in the UK and Denmark use RCT
evaluation designs and in the other countries evaluations involve quasi-experimental or observational
methodologies.
RAND OMIS ED C ONTR OL T RIA LS ( RCTS )
UK
In the mid-1990s new Early Years Centres were established in areas with high levels of deprivation in the UK. In
setting up one such centre in Hackney (a deprived borough in London), the Hackney Day Care Study proposed to
assess the effects of providing day care to children aged six months to three-and-a-half years from socially
disadvantaged families. Toroyan et al. (2003) attempted to implement a RCT study of ECEC and it took place in an
Early Excellence Centre, targeted on disadvantaged families. In practice it was a RCT of being allocated a place at
a particular Early Excellence Centre in Hackney, a socially mixed area including a high proportion of
disadvantaged families. The intervention group, chosen from a waiting list from which the control group was also
selected, were allocated a place at the Early Excellence Centre. Most of the control group used other forms of
day care.
The results found by the study are (1) an increase in the likelihood of mothers in the intervention group being in
paid employment, but with no increase in family income and (2) the intervention group children were more likely
to be infected with 'glue ear' (otitis media with effusion) probably because of the increased likelihood of cross-
child infections in group care. There were no child development effects and no positive cost benefits found but,
as the authors say, estimates were imprecise.
There are several points to be considered. The sample size is small. Using lenient criteria the authors' own
statistical power calculations indicated the need to recruit 140 mothers, whereas they actually recruited 120 with
only 51 being in the intervention group. This small imbalanced sample size reduces the power to detect
differences and makes the study vulnerable to chance variation. An illustration of this is the substantial
differences between the two groups at pre-test, despite the supposedly random method of selection. However,
these substantial differences were not studied as the authors came to this decision: ‘Statistical tests were not
conducted as this is not considered good practice’. These initial group differences make the likelihood of results
consistent with the intervention hypothesis more likely if children with higher pre-test scores are more likely to
improve and less likely if the opposite is the case. Also, to be a worthwhile RCT the control group would need to
have meaningful differences in exposure to day care. The fact that the majority of the control group in this study
used day care makes the study's value questionable. While it is impossible to predict with certainty whether the
control group would use day care, an inspection of the base rate of day care use in Hackney would have indicated
that the majority would do so. In addition, the procedures for assessing child development would not usually be
chosen by researchers experienced in child development, as several more sensitive procedures are available.
Overall this study provides little evidence of significance, due to its methodological limitations. However it does
613318 Deliverable 4.1
CARE Project Review of research on impact of ECEC
CARE: http://ecec-care.org/ - page 28
demonstrate the limitations of the RCT design for behavioural interventions where inadequate control is
available for control group experiences, as typically happens where people can exercise choice.
DENMARK
Using a RCT, Jensen, Holm, and Bremberg (2011) explored the effects of the Action Competences in Social
Pedagogical Work with Socially Endangered Children and Youth (ASP-program), which aims at improving all
children’s well-being and cognitive functioning and specifically the situation for socially disadvantaged children.
They found statistically significant effects in favor of the intervention group on all five Strengths and Difficulties
Questionnaire (SDQ) dimensions: emotional symptoms, conduct problems, hyperactivity/inattention, peer
relations and pro-social behaviour. More precisely, authors report that all indicators suggest positive effects from
the intervention, but only the effects for conduct problems and hyperactivity-inattention are statistically
significant and only at a ten per cent significance level, initially but later there are larger effects for both
emotional symptoms and conduct problems, which become significant at the one per cent level. A similar effect
for hyperactivity inattention is significant at the five per cent level. For peer relationship and pro-social behavior,
there were no significant effects. In sum, the intervention seems to have had a positive and growing effect (i.e.
negative parameters) on emotional symptoms, conduct problems and hyperactivity inattention, but not on peer
relationship and pro-social behavior.
QUASI- EXP E RIM E NTA L AND LON GITU DIN AL S TUDI ES
GERMANY
In Germany, the Socio-economic Panel (SOEP) survey of private households is a wide-ranging representative
study with annual follow-ups (DIW Berlin). The public kindergartens analysed in the SOEP primarily targeted four
and five year-olds and were designed to promote social and cognitive development, and were mostly available
on a half-day basis. Only about 20 per cent of all kindergarten places were full-day care. Hence working parents
usually need additional care arrangements such as neighbours, grandparents etc. By law, the German
kindergarten supports parents' labour market participation and helps parents meet family life responsibilities,
and it is seen as the first stage of the education system. Kindergarten is generally provided by the community or
non-profit organisations. It is intended to prepare children for school but is not compulsory. Kindergartens
receive high public subsidies and are supposed to be available for every child. In Germany, family day care is
typically used for toddlers (Spiess, Büchel, & Wagner, 2003). The main research question concerned the effect of
kindergarten attendance on the probability of later attending a school with extended academic requirements,
‘Gymnasium’, ‘Realschule’ or ‘Hauptschule’ (Burger, 2013; Landvoigt, Muehler, & Pfeiffer, 2007; Spiess et al.,
2003).
The SOEP data was used by Felfe and Lalive (2012). They employed a marginal treatment effects framework
(MTE), and found that children from advantaged families derived lower returns to childcare attendance than
children from a less advantaged family background. In particular, children from an advantaged background have
lower returns to childcare in terms of comprehension and independence (their ability to dress alone). One
explanation might be that higher socioeconomic status (SES) families invest already early on in their children’s
skills and thus, any benefits their children derive from childcare might not be captured by the measures used.
The results indicate that children who have the lowest returns from attending childcare are sent to childcare first.
Yet, children who would benefit the most- younger children and children from disadvantaged backgrounds- are
least likely to be sent to childcare.
The effects on language skills are particularly strong in regions with high migration rates. Children who are the
least likely to enter childcare (in terms of unobserved features) have higher gains from attending childcare in
613318 Deliverable 4.1
CARE Project Review of research on impact of ECEC
page 29 CARE: http://ecec-care.org/
terms of social, language, daily and motor skills than children who face lower unobserved entry barriers. Gains
are particularly high for younger children, boys, low birth weight children and children from low SES families.
NETHERL ANDS
Several early education and care programs were researched in the Dutch Cohort Study of Primary Education
(PRIMA) (Driessen, 2004). The common aim of these programs was to stimulate socio-emotional and cognitive
development. Various institutions targeted different age groups between birth and eight years, and were mostly
part-day. Day care centers provide childcare for children between birth and four years of age. They are generally
open every work day and usually administered by private organisations. The cost of day care centers is subsidised
by the government, based on parents’ income. Preschools or preschool playgroups target children between two
and four years and are available two to three half-days a week. They are financed by municipalities which usually
charge a fee to parents. ECEC programs, finally, are special services typically aimed at children from
disadvantaged backgrounds and usually conducted in preschools or elementary schools. The programs are
intended for children up to eight years of age. Three-quarters of programs are partly financed by municipal
authorities, and one-quarter by the ministry of welfare. The Dutch Public Preschool Study (DPPS) drew on public
preschools, (kindergartens), which are integrated in the primary school system, forming the first two grades of
primary school (van Tuijl & Leseman, 2007). Their curriculum is predominantly developmental: Most preschools
work with mixed-age groups; most time is spent in free-play activities and work lessons with children in small
groups. Whole group activities are regularly provided as start, break, or closing activities during the day and
include book reading, play, talking, and singing. In the second year of preschool, these activities are
complemented by literacy and math activities (exploring letters and words, counting, measuring etc.).
The PRIMA-cohort study starts with five year-old children attending kindergarten classrooms of primary schools.
Using retrospective analysis, reconstructing attendance to targeted preschool programs from age three to six and
controlling for family background characteristics based on school records, no statistically significant effects of
targeted preschool education were found on language and cognitive outcomes and school achievement
(Driessen, 2004). Similarly a recent retrospective analysis of targeted preschool and kindergarten programs for
three to six year-olds, with more detailed measures of the preschool programs, also did not find significant
effects of targeted programs (Bruggers, Driessen, & Gesthuizen, 2014). The retrospective approach in these
studies, however, is vulnerable to hidden (self-) selection mechanisms and lack of control for program
implementation. Other evidence (Slot, 2014) indicates that placement in targeted preschool programs is
selective. That is, children with the most cognitive and language delay are most likely to enter these programs,
regardless of socio-economic and ethnic-cultural background. Screening of language delays and referral to
targeted preschools is an official task of the infant and child public health care system in the Netherlands. Thus
the Dutch retrospective studies cannot provide reliable evidence.
van Tuijl and Leseman (2007), in a smaller scale prospective study, of Turkish-Dutch and Moroccan-Dutch four to
six year-old children (N=312) in the (20 hours/week) public kindergarten program of primary schools used age-
norm-referenced intelligence measures to examine growth and catch-up effects relative to age-norms that were
related to attendance of kindergarten. They reported gains with standardized effect sizes of 0.58 S.D. for total IQ,
and of 0.36 S.D. and 0.44 S.D. for verbal and fluid IQ respectively. Effect sizes were largest for the youngest non-
Dutch speaking children, entering the program at age four. In addition, gains in verbal and fluid intelligence
predicted receptive vocabulary and pre-mathematical skills at age six, with the strongest effects for the youngest
non-Dutch speaking children.
de Haan, Elbers, Hoofs, and Leseman (2013) followed two cohorts of preschoolers of low-income immigrant
families (N = 91), aged three and four years at the start of the study, attending preschools and kindergarten
classrooms in primary schools. Children were attending classrooms with or without an educational program
intended to increase language and pre-math skills, and classrooms with a socioeconomically mixed or mainly
disadvantaged composition. Classroom observations determined the amount of time spent on language, literacy
and math activities initiated by the teacher. Using a cohort-sequential augmented latent growth analysis, the
study showed positive effects of teacher-initiated language, literacy and math activities on children’s growth in
these skills over time, with effect sizes (standardized regression weights) ranging from 0.35 S.D. to 0.52 S.D. for
teacher initiated academic activities and from 0.30 S.D. to 0.62 S.D. for classroom composition. Interestingly,
613318 Deliverable 4.1
CARE Project Review of research on impact of ECEC
CARE: http://ecec-care.org/ - page 30
working with or without an education program did not explain the variance in teacher initiated activities, nor
affect children’s growth in academic skills, pointing to weak program implementation.
Blok, Fukkink, Gebhardt, and Leseman (2005) conducted a meta-analysis comparing targeted centre
and home-based programs for disadvantaged children for the age range zero to six years. Centre-
based programs and combined centre-based and home-based programs were more effective by about
0.5 S.D. on cognitive and educational outcomes than home-based programs, but they found no
consistent effects for socio-emotional outcomes.
FRANCE
The French kindergarten, the ‘ecole maternelle’, is available to all children from three to six years and it has an
explicit educational mission although not all necessarily focus on the promotion of pre- academic skills (some
primarily emphasise the promotion of social development instead). The French kindergarten is fully funded by
the State as is part of the national education system and attended by almost all three to five year-olds (OECD,
2006).
While this is a universal programme its possible benefits for disadvantaged groups have been separately
investigated. Caille (2001) reported a stronger effect of an earlier start in ‘ecole maternelle’, at age two compared
to age three, on early school skills and grade retention in primary school, especially for low-income and
immigrant ethnic minority children. Also, it has been shown (Dumas & Lefranc, 2010) that the large-scale
expansion of a universal, free preschool program led to nearly universal preschool attendance in three and four
year-olds and this appeared to reduce socio-economic inequalities as children from less advantaged backgrounds
benefitted most.
S U M M A R Y O F E V I D E N C E F O R D I S A D V A N T A G E D
C H I L D R E N
Many of the interventions for disadvantaged groups have involved high quality ECEC. The evidence on childcare
in the first three years for disadvantaged children indicates that high-quality ECEC can produce benefits for
cognitive, language and social development. There are indications that low-quality childcare may not produce
such benefits. High-quality childcare with associated home visits may produce the greatest benefit for children
less than three years old.
With regard to provision for three years onwards disadvantaged children benefit particularly from high-quality
preschool provision. Also children benefit more in socially mixed groups rather than in homogeneously
disadvantaged groups (Melhuish et al., 2008). Some interventions have shown improvements in cognitive
development, and in some cases these persist throughout children's school careers. However, early childhood
interventions do boost children's confidence and social skills, which provides a better foundation for success at
school (and subsequently in the workplace). Reviews of the research often infer that it is the social skills and
improved motivation that lead to lower levels of special education and school failure and higher educational
achievement in children exposed to early childhood development programs. However there is clear evidence
that cognitive, language and academic skills can also be enhanced by ECEC experience and these are likely to
play a role also in the later educational, social and economic success that is often found in well-implemented
ECEC interventions. Studies into adulthood indicate that the educational success is followed by increased
613318 Deliverable 4.1
CARE Project Review of research on impact of ECEC
page 31 CARE: http://ecec-care.org/
success in employment, social integration and sometimes reduced criminality. There is also an indication of
improved outcomes for mothers. The improvements appear to occur for those problems that are endemic for
the particular disadvantaged group, and hence show the greatest room for improvement.
P A R T 2 : E C E C F O R T H E G E N E R A L P O P U L A T I O N
A T T E N D A N C E P A T T E R N S O F C A R E
Children’s attendance patterns at early education and care settings, and the age at which they first attend, can
vary greatly. Understanding the individual and combined effects of variations in these factors is important for
both researchers and policy makers; yet in reality they are hard to disentangle. Timing, duration, and intensity of
ECEC attendance can be highly correlated. For example, in the US, the majority of children receive care in their
first year, and once care is initiated, the amounts of time spent in an ECEC setting stay high and stable
throughout early childhood (Bradley & Vandell, 2007; Phillips & Lowenstein, 2011). Thus, findings on the effect of
ECEC attendance patterns on developmental outcomes generally reflect combined effects of timing and amount.
However complex patterns involving amount, type and timing of care do occur in some countries as, for example,
reported in the UK by the ‘Families Children and Child care’ (FCCC) study (Eryigit-Madzwamuse & Barnes, 2014).
Such differences between countries are but one reason why conclusions about ECEC effects may vary from
country to country.
A T T E N D A N C E A N D Q U A N T I T Y O F C H I L D C A R E
ECEC FOR C HILD REN AGED ZE RO TO THRE E YE ARS
Findings on the relationships between attendance or amount of early childcare and children’s development are
mixed: negative effects, no effects and positive effects (particularly for children at risk) have been identified (see
for example, Anders, 2013; Bradley & Vandell, 2007; Melhuish, 2004b; Ruzek, Burchinal, Farkas, & Duncan, 2014;
Vandell, 2004; Zaslow et al., 2010).
613318 Deliverable 4.1
CARE Project Review of research on impact of ECEC
CARE: http://ecec-care.org/ - page 32
SOCIO- E M OT I O NAL DEVELOP M E NT
The consequences of day care for socio-emotional development, and especially attachment and behaviour
problems, are extremely contentious, because of the strong emotional reactions aroused.
ATTACHMENT
The issue of whether day care is bad for children was partly derived from the theoretical work of Bowlby (1951,
1969) on the development of an attachment by the infant towards the principal caregiver, usually the mother.
Attachment has come to be seen as a fundamental step in development and that disruption to attachment may
have longer-term developmental consequences (Ainsworth, Blehar, Waters, & Wall, 1978). A child’s attachment
to the mother may be classified as secure or insecure, with secure attachment leading to positive development
but insecure attachment being associated with an increased risk of negative developmental outcomes. Examples
of developmental sequelae are; children with secure relationships to their mothers when infants have been
reported to be more sociable (Pastor, 1981), and more socially competent in preschool (LaFreniere & Sroufe,
1985; Waters, Wippman, & Sroufe, 1979). The perspective that daily separations may reduce secure attachment
influenced much early research on infant childcare.
Two 1980’s meta-analyses summarised many US studies and concluded that non-maternal care in the first years
of life could increase the likelihood of insecure attachment with the mother (see Friedman & Boyle, 2008, for a
comprehensive review of day care and attachment in the US). Concerns possible effects of non-parental care
upon infant attachment, as well as concerns over sample sizes and the quality of previous studies were
instrumental in the setting up of the hallmark study known originally as the National Institute of Child Health and
Human Development Early Child Care Research Network (NICHD ECCRN) and now known as the NICHD Study of
Early Child Care and Youth Development (NICHD SECCYD), comprising approximately 1300 children sampled from
ten sites across the US.
The NICHD Study of Early Child Care and Youth Development (NICHD SECCYD) is a landmark study and deserves
detailed consideration. It is paid more attention and given greater credibility for its results than other studies.
Certainly it has spent more money per child, had more researchers per child and more measurements per child
than any other study in this field. However it has limitations. The sample size is only 1300 and substantial
amounts of data are missing across the study period. Hence reports of the study are largely reliant upon results
from analyses of imputed data. Typically imputation methods assume data are missing at random (MAR),
meaning that the nature of missingness is related to other measured variables and these can be used in the
imputation model to make reasonable estimates of missing values. However there is reason to believe that the
missing data on quality of childcare in the NICHD-SECCYD study is missing not at random (MNAR), i.e. it is likely
that refusal to participate in quality measurement is more likely for the lowest quality settings, and hence the
value of the quality variable is linked to missingness and other measured variables do not account for
missingness. All methods for incomplete data imputation under MNAR make unverifiable assumptions, because
it is very rare to know the appropriate model for the missingness mechanism. Hence imputation under these
circumstances is inherently open to question. In these circumstances (i.e., lower quality care more likely to be
missing) it is likely that estimates based on data imputed under MAR assumptions will underestimate or miss any
effect related to the quality of childcare.
Regarding US studies it should be noted that the nature of employment rights, parental leave and childcare
availability in the US mean that parents who wish to return to employment after the birth of a child and use
childcare have typically done so from the time when the child is a few months old. They will then usually
continue to be employed and use childcare. This results in a situation where those children who receive childcare
in the first year are likely to be largely the same as those receiving childcare in the second and third years so that
amounts of time in ECEC remain stable throughout early childhood (Bradley & Vandell, 2007; Phillips &
Lowenstein, 2011). The consequence is that there are limits to the investigation of age of starting and timing
effects for childcare. Hence analyses of quantity of childcare become similar to analyses of the intensity of
613318 Deliverable 4.1
CARE Project Review of research on impact of ECEC
page 33 CARE: http://ecec-care.org/
childcare (hours per week) as the variation due to age of starting has less influence on the quantity of childcare
measure.
With regard to measuring the quality of childcare, the study used several measures of process and structural
quality. However a substantial number of childcare settings used by children in the study did not cooperate with
the quality assessment visits. It is possible, even likely, that the settings that did not cooperate were settings
where the quality is lower than in other settings. To the extent that this occurs, the analyses will underestimate
possible effects due to variation in quality. Additionally, if the characteristics of the missing quality measurements
are not reflected in settings for which quality measurements are available then imputation will only replicate the
range of variation in the complete data and hence even analyses of imputed data will underestimate the
potential effects of quality of childcare variation.
Contrary to expectations, this study showed no main associations between either quantity (including type of care
and age of entry) or quality of care, and attachment security with the mother at ages 15 and 36 months,
measured with the Strange Situation (Friedman & Boyle, 2008). However, if quantity was high in the first 15
months (something that now rarely happens in countries with extensive parental leave such as Norway or the
UK) and also either day care was of low-quality or unstable, or parental sensitivity was low, only then was the
likelihood of insecure attachment somewhat increased. Other studies also find that higher rates of insecure
attachment are associated with poor quality daycare (Harrison & Ungerer, 2002; extended analyses in Love et al.,
2003). In summary, daycare may compromise attachment security, but only in instances of poor quality infant
care either at home and/or in daycare. This explanation gets support from findings showing that more hours of
non-maternal childcare in the first three years can relate to more negative and less positive parenting (Belsky,
1999) or less maternal sensitivity (NICHD Early Child Care Research Network, 1999a). On the other hand, higher
quality non-maternal care predicts greater maternal sensitivity (NICHD Early Child Care Research Network,
1999a).
The timing of childcare in terms of child’s age can be important. Varin, Crugnola, Molina, and Ripamonti (1996)
looked at the effects of age of entry into childcare on the development of attachment amongst 129 children.
Their analysis was based on care-givers evaluating the behavior of the children using the Day Care Adaptation
Scale at three years of age. They also observed the reunion between 54 of the children and their parents. They
found that children who started childcare between six to 12 months and 18 to 23 months were more prone to
frustration and had difficulty reuniting with their mothers. In contrast, children who started between 12 to 17
months displayed lower levels of relational distress.
Furthermore, infants and toddlers, securely attached to their primary caregivers, may find experiencing and
settling into day care less stressful. This hypothesis gets support from a German study (Ahnert, Gunnar, Lamb, &
Barthel, 2004), which showed that, during their adaptation to day care, infants classified as less securely
attached had higher cortisol levels, and showed more negative affect. Importantly, attachment security to the
parent was related to the time spent adapting their children to day care - if more time was spent, attachment
remained secure or became secure.
Children may also form attachment relationships with caregivers in ECEC. Secure attachment with caregivers is
less frequent than with parents (40% vs. 60%), and caregivers showing high levels of sensitive responsiveness
were more likely to have children securely attached to them (Ahnert, Pinquart, & Lamb, 2006).
OTHER ASPECTS OF SOCIAL AND EMOTIONAL DEVELOPMENT
Other reports have focused on subsequent aspects of social and emotional development, such as behavior
problems and social skills. Quantity of group care, in particular where there is early age of entry and high hourly
amounts have been associated with somewhat elevated levels of externalizing behaviour problems in the SECCYD
NICHD ECCRN study, as in other US studies (e.g. Loeb, Bridges, Bassok, Fuller, & Rumberger, 2007) and other
countries (Belsky, 2001; Yamauchi & Leigh, 2011). Sometimes associations remain when outcomes are measured
in later childhood and adolescence (see Zachrisson, Dearing, Lekhal, & Toppelberg, 2013, for review). However,
while earlier reports from the NICHD SECCYD found that negative effects on externalizing behaviours were
independent of ECEC quality, emphasising quantity and type of care (Belsky et al., 2007), later analyses from the
613318 Deliverable 4.1
CARE Project Review of research on impact of ECEC
CARE: http://ecec-care.org/ - page 34
NICHD SECCYD have found that quality does moderate the effects of quantity and type of care on behavioural
outcomes (McCartney et al., 2010). This study found that high-quality childcare could partially compensate
negative behavioural effects of high quantity childcare. A recent paper that aimed to get a better understanding
of the links between quantity of childcare and children’s behaviour problems and poor social skills, reviewed
results from the largest and most representative childcare studies in the US as well as many other smaller
studies. It concluded that there is a real effect of childcare hours on children’s behaviour and good quality care
does not completely eliminate these effects. However, effects are context specific (occurring primarily for
teacher/caregiver reports), specific to particular aspects of social behaviour (particularly externalizing behaviour
problems), specific to populations not facing disadvantage, and a function of cumulative hours rather over time
rather than any particular developmental period (Huston, Bobbitt, & Bentley, 2015).
A longitudinal study in Quebec (Pingault et al., 2015) found that children who received non-parental child-care in
the pre-school years were rated by teachers as less shy, less socially withdrawn, more oppositional, and more
aggressive at school entry (age 6 years). However, these differences were related to greater amounts of group
care experience and disappeared during elementary school as children who had received exclusive parental care
caught up with those who received preschool group care. For oppositional and aggressive behaviours these
differences had disappeared by 8 years of age. Similar results relating to antisocial behaviour to group care in the
first two years of life had previously been reported for an English longitudinal study by Melhuish (2010) where
initially more antisocial behaviour was associated with more group care in the first two years of life but this effect
gradually reduced and then disappeared during the elementary school years, Similar results occurred in a parallel
study in Northern Ireland (Melhuish, Taggart, Siraj-Blatchford, & Sammons, 2006). In another British study of
childcare use at three, ten, 18 and 36 months, Barnes et al. (2010) failed to find an association between amount
or type of childcare and mother-reported levels of disruptive behaviour at 36 months of age. It would appear
that any link between group care and later externalizing or antisocial is not as persistent or long-term in non-US
research.
Studies in socio-political contexts with more stringent quality regulations and greater parental leave than in the
US also question whether these negative associations are universal (Love et al., 2003). For instance, in two
studies in Norway, Solheim, Wichstrøm, Belsky, and Berg‐Nielsen (2013) and Zachrisson et al. (2013) failed to find
associations between quantity of care and externalizing behavior, the latter using statistical methods that mimic
the effects of randomisation. Furthermore, Lekhal (2012) using data from the Norwegian Mother and Child
Cohort Study (MoBa; data also used in Zachrisson et al., 2013), found no associations between the use of
childcare in the first three years of life and externalizing and internalizing problems. Moreover, using data from
the Behaviour Outlook Norwegian Development Study (BONDS), Dearing, Zachrisson, and Naerde (submitted)
examined the links between age of entry into continuous ECEC, both extensive and not, and the development of
aggression across ages two, three and four years. Based on teacher reported aggression, they found that early
entry into ECEC was associated with modest increases in physical aggression at age two, however, by age four the
effect of age of entry disappeared, while statistical techniques mimicking randomisation failed to confirm even
this early association.
While Zachrisson et al. (2013), did not find any association between quantity of childcare and externalizing
behavior, they did find some evidence that larger group size in childcare may be associated with increased
externalizing behavior, and less social competence.
Nonetheless, two Norwegian studies have found associations between age of entry and behavioural problems.
Schjølberg, Lekhal, Vartun, Saugestad Helland, and Schjelderup Mathiesen (2011) found that childcare before 18
months of age was slightly correlated with both language difficulties and behavioural problems at age five.
However, this study did rely on less stringent methodology, with poor control of selection effects, than the other
Norwegian studies cited. Hence their results may reflect methodological limitations. A study that collected data
around 30 years earlier in Norway (Borge & Melhuish, 1995) did find a link between anti-social behaviour and
amount of early childcare. The differences in results between this very early study in Norway and recent studies
may well be a consequence of the extensive development of the childcare system in Norway over the past 30
years. In particular, when the earlier study data were collected parental leave was more limited so childcare in
the first year was common and the structural quality in terms of staff characteristics etc. was less well developed,
so that the quality of childcare was probably not as good as for the later studies.
613318 Deliverable 4.1
CARE Project Review of research on impact of ECEC
page 35 CARE: http://ecec-care.org/
Also, in the Netherlands, Broekhuizen, van Aken, Dubas, and Leseman (2014), it was found that high quantities of
childcare was linked to more teacher-rated externalizing behaviour both concurrently at age two, and one year
later at age three. However, for parent-rated externalizing behaviours, more days in high-quality childcare were
related to less externalizing behaviour. Follow-up analyses showed that the difference in externalizing behaviour
for children in relatively low or high-quality childcare became significant when children spent three days or more
in this high-quality childcare setting.
When comparing studies across countries, it appears that most consistent negative behavioural effects are found
for teacher reports (e.g. Broekhuizen, van Aken, Dubas, & Leseman, 2014; Loeb et al., 2007; NICHD Early Child
Care Research Network, 2006), and for families from relatively high SES families (Broekhuizen, van Aken, Dubas,
& Leseman, 2014; Loeb et al., 2007; Yamauchi & Leigh, 2011). Thus, mixed findings for effects of childcare
quantity appear to be partially due to variations in study designs, such as differences in informants and sample
characteristics. Moderation of ECEC effects by sample characteristics is discussed later in this review.
Like the study of McCartney et al. (2010) and Broekhuizen, van Aken, Dubas, and Leseman (2014), several other
studies have examined interactions between early childcare quantity and quality in relation to children’s socio-
emotional behaviours. For indicators of childcare structural quality, it was found that negative behavioural
associations with extensive time in childcare could be partially compensated by a low child-staff ratio (Yamauchi
& Leigh, 2011), and a low proportion of time spent in a large group of peers (McCartney et al., 2010). For
indicators of childcare process quality, it was also found that high process quality care could mitigate the
association between extensive time in childcare and child externalizing behaviour (McCartney et al., 2010).
Moreover, like the study of Broekhuizen, van Aken, Dubas, and Leseman (2014), Votruba‐Drzal, Coley, and
Chase‐Lansdale (2004) found that children who spent extensive time in high process quality childcare showed
the least mother-rated externalizing behaviour (Votruba‐Drzal et al., 2004). So while some studies found that
negative behavioural effects of high quantity care could partially be compensated by high-quality childcare
(McCartney et al., 2010; Yamauchi & Leigh, 2011), other studies found that positive behavioural effects of high-
quality childcare were even strengthened by spending many hours in this high-quality setting (Broekhuizen, van
Aken, Dubas, & Leseman, 2014; Votruba‐Drzal et al., 2004).
COGNITIVE, LANGUAGE AND EDUCATIONAL DEVELOPMENT
Overall, however, the studies in this review suggest positive effects of ECEC attendance under the age of three
with regard to children’s cognitive and language development, and their academic achievement – some effects
lasting into later childhood and adolescence. This seems particularly true for attendance in centre-based care,
and for children starting to attend ECEC settings between the ages of two to three years (see also recent reviews
by Leseman, 2009; Mathers, Eisenstadt, Sylva, Soukakou, & Ereky-Stevens, 2014). Furthermore, centre-based
care at an early age seems to be particularly beneficial for children from educationally disadvantaged families.
However, there is strong evidence that not only the disadvantaged, but also the general population, benefit from
preschool education (Melhuish, 2011).
US Studies
An early example of a community-based approach to ECEC was the Brookline Early Education Project (BEEP). This
project began in 1972 providing a range of health and centre-based care and education services from birth to
school to families in the intervention group. The intervention was open to any family in Brookline, Boston. At
both kindergarten and third grade BEEP children scored higher and demonstrated fewer difficulties in social
development and learning skills than comparison children from the same classrooms and similar family
backgrounds. Participation in the highest level of program services, for instance, tended to close the
performance gap between children of more educated and less educated mothers (Theroux, 2006). As young
adults the intervention group reported higher incomes, less depression, better employment, better health and
less risk-taking behaviour than the comparison group (Hauser-Cram, Pierson, Walker, & Tivnan, 1991; Tremblay,
Pagani-Kurtz, Mâsse, Vitaro, & Pihl, 1995). Long-term follow-up revealed that for disadvantaged students,
differences between those enrolled and not enrolled were evident as late as age 25 (Herrod, 2007; Palfrey et al.,
2005).
613318 Deliverable 4.1
CARE Project Review of research on impact of ECEC
CARE: http://ecec-care.org/ - page 36
The US NICHD Early Child Care Research Network study of early childcare on developmental effects of day care in
ten US communities brought strong evidence that more experience of centre-based ECEC predicted better
cognitive and language outcomes at 24, 36, and 54 months (NICHD Early Child Care Research Network, 2000b,
2002b, 2003a, 2004, 2005b). By third grade, however, centre-care exposure predicted only enhanced memory,
but no longer superior academic achievement (Belsky et al., 2007; NICHD Early Child Care Research Network,
2005b). Advantages of more centre-based experience in early childhood were not found to persist into
adolescence (NICHD Early Child Care Research Network, 2002b; Vandell, Belsky, Burchinal, Steinberg, &
Vandergrift, 2010). Additionally, the report on this study by Vandell et al. (2010) shows the moderating effect of
quality on age 15 cognitive effects, with standard effect sizes of d = 0.14 to 0.19 on cognitive academic
achievement for the highest quality, but effects were very small to very small negative for low to moderate
quality.
Data from the US birth cohort Early Childhood Longitudinal Study showed that – overall – centre-based care
raises reading and math scores, and that the intensity of centre-based care matters: more hours per day leads to
greater benefits (Loeb et al., 2007). Using data from 9,185 children (five years and older) who participated in
another US cohort study (Children of the National Longitudinal Survey of Youth), it was reported that those
children starting non-maternal care in the first two years had higher achievement scores in childhood and
adolescence compared to those children not attending ECEC under the age of three (Jaffee, Van Hulle, &
Rodgers, 2011). Data from 317 US children enrolled in kindergarten found that months in ECEC during the infant,
toddler and preschool period predicted letter recognition skills (Christian, Morrison, & Bryant, 1998). And a study
of 229 US children enrolled in childcare at the age of two-and-a-half showed positive effects of centre
attendance on children’s cognitive skills at age seven-and-a-half (Bassok, French, Fuller, & Kagan, 2008).
Alongside a body of strong evidence of beneficial effects of early ECEC attendance, there are some
research findings that point to no effects, or negative effects (Bornstein, Hahn, Gist, & Haynes, 2006;
Driessen, 2004; Lefebvre & Merrigan, 2002). Jaffee et al. (2011), for example, reported for their US
cohort sample (Children of the National Longitudinal Survey of Youth) positive effects for between-
family comparisons, however when comparing siblings within the same family, different experience of
early care had neither positive or negative effects on children’s outcomes.
Early results from the NICHD study showed that children without early non-maternal care experience showed no
differences to those children with early non-maternal care experience (NICHD Early Child Care Research
Network, 2002b, 2002c), and that the quantity, or average amount of time children spend in ECEC each week,
was not related to children’s cognitive skills, language skills, or to their school readiness prior to school entry.
Children did not gain any greater benefit from spending more time in childcare, even in childcare rated as being
of high-quality (NICHD Early Child Care Research Network, 2003a). Yet time, in certain types of care (mainly
centre-based) was shown to be effective for children’s language and cognitive development (NICHD Early Child
Care Research Network, 2000b, 2002b, 2003a, 2004, 2005b).
Results from the Early Childhood Mental Health Program showed that hours of non-maternal care across infant,
toddler, and preschool age were not predictive of children’s language and cognition at 54 months (Bornstein et
al., 2006).
Australasian studies
The New Zealand study ‘Competent Children, Competent Learners’ Study (Wylie & Thompson, 1998)
demonstrated that attending centre-based care before the age of three was related to benefits for children’s
cognitive-linguistic development. While some effects on children’s academic skills lasted into adolescence (Wylie,
Hodgen, Ferral, & Thompson, 2006), an analysis at age 16 showed that effects were no longer visible (Hogden,
2007; Wylie, Hodgen, Hipkins, & Vaughan, 2009).The Longitudinal Study of Australian Children reported better
learning outcomes at age six years for children attending up to 24 hours of ECEC per week. Again there are
studies indicating possible negative effects. The Sydney Family Development Project reported generally positive
effects of early ECEC attendance; however, they also found negative effects for children’s learning at age six in
combination with many hours within the first three years (Love et al., 2003).
613318 Deliverable 4.1
CARE Project Review of research on impact of ECEC
page 37 CARE: http://ecec-care.org/
European and international studies
In the UK, the Effective Pre-School, Primary and Secondary Education Project (EPPSE) study found that children
who attended centre-based care before the age of three years (and in some cases before the age of two) had
better cognitive and language skills at age three and just prior to school entry (Sammons et al., 2002; Sylva,
Melhuish, Sammons, Siraj-Blatchford, & Taggart, 2010). The English FCCC study found that more hours of group
care up to 18 months related to higher cognitive scores at 18 months, with a marginal effect on language
development (Sylva, Stein, Leach, Barnes, & Malmberg, 2011).
In Sweden, researchers found that longer duration of ECEC attendance before the age of four was
related to enhanced cognitive and language development. Advantages persisted into later childhood
(Broberg, Hwang, Lamb, & Bookstein, 1990; Broberg, Wessels, Lamb, & Hwang, 1997), but could not
be found any more in adolescence or adulthood (Hwang, 2006). In Norway, Dearing, Zachrisson,
Mykletun, and Toppelberg (submitted) found that ECEC at age one-and-a-half had positive effects on
early language, in particular for children from lower and middle-income families, providing causal
estimates of the effects of ECEC.
In Germany, a national ECEC study did not show differences between those children with and without ECEC
experience in terms of their linguistic and cognitive skills. However, at two time points (age two-and-a-half and
age four) the study found evidence that an earlier starting age related to better language skills (Tietze et al.,
2013). Also it was reported that attendance in ECEC before the age of three increased the likelihood of attending
a more academically focussed secondary school system, offering final pre-university exams, from 36 per cent to
50 per cent (Fritschi & Oesch, 2008). Again in Germany, Lalive (2010) found that ECEC attendance in the first
years increased language skills among two to three year-old children, and school grades in middle childhood. In
Italy, the availability of additional childcare slots for the zero to two age group was shown to be effective in
raising children’s language scores, but not for their maths skills (Brilli, 2012; Brilli, Del Boca, & Pronzato, 2014).
In the Netherlands, Luijk et al. (2015:1) recently looked at the association between children’s language
development, between the ages of one and six, and the amount of hours they spent in non-parental care. They
found that in general ‘more hours in non-parental childcare were associated with better language abilities.
However, more hours in care in the first year of life were associated with less language proficiency at ages 1 to
1.5’ Additionally, they also found that children who attended more hours of centre base care, compared to