Outcomes of early childhood education: Literature review
This literature review was commissioned by the Ministry of Education to provide policy makers with a synthesis of research that analyses the impact of early childhood education (ECE) for children and families.
Author(s): Linda Mitchell, Cathy Wylie and Margaret Carr, New Zealand Council for Educational Research.
Date Published: May 2008
This report is available as a download (please refer to the 'Downloads' inset box). To view the individual chapters please refer to the 'Sections' inset box.
Chapter 2: Children - Cognitive outcomes
"Cognitive" outcomes are the focus of much of the research on the impact of ECE. Cognitive outcomes were mainly defined as "academic knowledge" based on traditional academic subjects, or ability as measured by IQ tests or developmental quotients. Measures used were achievement tests or teacher assessments of school performance, particularly in mathematics, reading, and literacy; IQ; school readiness; grade retention; and special education placement. These outcomes are reported separately in many studies and in this chapter, but they are also linked to learning dispositions and social-emotional outcomes. Indeed, the longitudinal case studies of practice in the English EPPE study of over 3000 children and 141 centres found "The settings that viewed cognitive and social development [including learning dispositions] as complementary seemed to achieve the best outcomes" (Siraj-Blatchford et al., 2003, p. vii).
In this and subsequent chapters we move on from our review of pre 1995 studies in Chapter 1, to examine research reports from 1995 onwards. The studies used are four U.S. intervention studies chosen because they followed children long term (the Abecedarian study, the High/Scope Perry Preschool study, the Chicago Child-Parent Centre study, and the Infant Health and Development Program study), studied general everyday ECE experiences, and studies using national survey data.
In this section, we have mapped out the size of results for cognitive outcomes, highlighted consistencies and differences across studies, and then analysed differences associated with child and ECE characteristics. Effect sizes are reported after taking account of background variables unless stated otherwise.
In all, 26 studies reported on mathematics outcomes from ECE participation. Those that provided greatest insight for our research questions examined quality of ECE, types of ECE, stability, timing and duration of ECE, and characteristics of children.
Most studies comparing children who participated in ECE with those who did not, found positive gains from ECE participation for mathematics at the time of attendance and in the early years of schooling. In general, the intervention studies reported medium to large effect sizes at age 8, and the general studies reported small to medium effect sizes. All six studies measuring long-term effects of ECE participation found benefits for children continuing at least until at ages 15-16.
One study found negative impact of centre-based ECE for children from low-income families, but positive impact for children from high-income families.
Three studies found no impact of ECE on mathematics outcomes.
Consistent benefits from Abecedarian, Chicago Child-Parent Centre and Perry Preschool programmes (small to medium effect sizes d=0.29-0.45)|
No difference in 1 study (IHDP), but benefits for heavier birth-weight premature children
|Consistent benefits from Abecedarian, Chicago Child-Parent Centre and Perry Preschool programmes (small to medium effect sizes d=0.19-0.44)|
|General Everyday ECE studies||Consistent benefits in 7 studies (small to medium effect sizes d=0.18-0.43)||Consistent benefits in 1 study, benefits from good quality ECE in Competent Children, Competent Learners study at age 14, but not 16|
|Studies using National Survey Data||
Benefits in 5 studies|
Mixed impact in 1 study (benefits for children from high-income families, children from low-income families did less well)
No impacts for children attending Head Start with peers who did not in 2 studies; no greater impact for children attending higher-quality centres compared with those attending poorer-quality centres in 1 study
|Consistent benefits in 2 studies (but not significant in 1 of these—small effect size d=0.05)|
Studies of the Abecedarian, Chicago Child-Parent Centre and Perry Preschool programmes reported medium to large effect sizes at age 8 (d=0.29-0.81), for the full intervention versus none. The Infant Health and Development Programme, which did not find benefits overall, did find benefits for the heavier but not the lighter birth-weight premature children at age 8. The authors suggest the lighter birth-weight children may have needed continued support (the intervention finished at age 3), or may have included more neurologically impaired children who could not benefit from the intervention.
U.S. studies, the English EPPE study and Northern Ireland EPPNI study that followed children through from ECE programmes to the third year of school reported small to medium effect sizes (d=0.07-0.43), with most in the range d=0.23-0.29. Many of these are U.S. studies. The lowest effect sizes were in Gilliam and Zigler's (2004) summary of U.S. pre-K evaluations in the U.S. (using comparisons with national norms or comparison groups who may or may not have attended ECE), and a study of an entire kindergarten cohort in a large urban U.S. setting (Fantuzzo et al.., 2005), comparing those with centre-based ECE with those who had experienced informal care or no extra care. For example, in the pre-K evaluations, small effect sizes reported for South Carolina first grade and Texas 3rd grade ranged from d=0.07 to .09. Michigan reported an increase of 16 percent more students passing the Michigan Educational Assessment Mathematics test at fourth grade. The highest effect sizes were medium, and were found in the EPPE study (d=0.43) and Oklahoma's pre-K programme (d=0.38) at school entry. In Oklahoma, some structural features of quality, i.e. teacher qualification requirements and pay rates, tend to be higher than in other states.
The EPPE study reported reducing effect sizes on early number from d=0.43 at school entry, to d=0.38 at age 6, and d=0.2 at age 7.
Studies Using National Survey Data:
Studies using U.S. national survey data reported small positive effect sizes in the first year of school (d=0.10 to 0.19). In Argentina, Berlinski, Galiani, and Gertler (2006) found positive impacts of pre-primary education on third grade standardised mathematics tests, following Argentina's expansion of universal pre-primary education between 1993 and 1999. They estimated one year of preschooling increased average third grade performance in mathematics (and Spanish) by 8 percent of a mean of 61.4 or by 23 percent of the standard deviation of the distribution of test scores. The Argentina preschools offered a curriculum aimed at developing personal autonomy and behavioural skills, social skills, logical and mathematics skills, and emotional skills, and average class sizes of 25 students and two shifts (i.e. 50 children in total).
Six studies reported long-term gains in mathematics performance, with effect sizes generally decreasing over time. The Competent Children, Competent Learners study, focusing on aspects of ECE quality, found advantages of some aspects of high-quality ECE at age 14 and age 16.
The Abecedarian project found medium effect sizes for mathematics of d=0.45 at age 8, d=0.35 at age 12, and d=0.44 at age 15. The Chicago Child Parent-Centre programme found smaller effect sizes of d=0.32 at third grade, d=0.24 at fifth grade, and d=0.19 at eighth grade (Karoly et al.., 2005). Both programmes combined parent education with good quality ECE, but the Abecedarian programme began in the child's first three months and the Chicago Child Parent-Centre programme at ages 3-4.
The Competent Children, Competent Learners study found modest contributions from early childhood education to age-14 and age-16 mathematics scores, with ECE quality variables and length of experience accounting for 2.5-5 percent of the variance in age-14 mathematics scores, and an ECE quality variable (ECE staff guiding the children) accounting for 4 percent of the variance in age-16 mathematics scores.
Two of the evaluations of pre-K summarised by Gilliam and Zigler (2004) measured long-term impacts. New York-EPK found significant positive impacts in mathematics at 6th grade. In Maryland there were statistically significant effects of pre-K participation on mathematics in 5th, 8th, 9th and 10th grades.
Studies Using National Survey Data:
Goodman and Sianesi, (2005), using British National Child Development Survey data found a small average gain for those who had attended ECE in mathematics tests of d=0.08 effect size at age 7, reducing steadily to d=0.05 effect size (not significant) at age 16.
Results for age-15 mathematics performance in the recent international PISA study showed students who had attended ECE for at least a year before school scored 8 points higher on average than those who had not, after taking socioeconomic background into account (OECD, 2004).
No ECE impact for mathematics was reported in three U.S. studies with children from low-income homes. Two found no gain for mathematics for children who had experienced Head Start, compared with their peers, at ages 3-4 (U.S. Dept. of Health and Human Services, 2005), and ages 12-17 (Aughinbaugh, 2001). However, it is not clear how different the two groups were, since the non-Head Start group included some who had other ECE experience. One study comparing differences within everyday ECE experience of quality and type found that those who attended comparatively higher-quality ECE had much the same mathematics scores as those who attended lower quality (Votruba-Drzal, Coley & Chase-Lansdale, 2004). However, in this study, most of the ECE settings were minimally adequate in meeting basic developmental needs, even those that were of higher-quality.
Currie and Thomas (2000) have shown that poor school quality can undermine early gains. Another explanation for why the Auginbaugh study did not show long-term gains for Head Start children is that they may have gone on to attend poor quality schools with disadvantaged peers. Votruba-Drzal et al. suggest reasons why their findings differ from other studies are that within the normative range of child care quality available to low-income children in their communities, the sample children may need higher-quality ECE than was found in the study. Even relatively high-quality child care may not be able to make up for other environmental challenges. These children may also need consistently high-quality experience over a longer period to gain cognitively. The study did not collect data on length of ECE experience, and the authors noted other studies, e.g., the NICHD study, showing gains for cognitive performance related to ECE length. The data used was only a snapshot of children's child care experiences.
One U.S. study using national survey data (Gamoran et al.., 1999) found that children from low-income homes who participated in centre-based care did less well on tests of mathematics, and high-income children did better, than similar children at age 6 who did not participate in this type of care, but could have been in non-maternal home-based or maternal care. Their study included only children with siblings and did not analyse centre care quality or duration. Burchinal and Nelson (2000) have summarised U.S. studies showing that children from more advantaged families tend to attend higher-quality child care than children from less advantaged families. Such selection factors could account for this study's findings. As well, a hierarchical model rather than the fixed effect model used may have been a more appropriate analysis for this data.
Differences for population groups
Gains in mathematics, from ECE participation were found for children across the board. However, there are also additional gains for children from low socioeconomic homes.
The Competent Children, Competent Learners study found an indication that children who had been in very low-income homes at age-5 appeared to benefit at age 14 more than most others in cognitive competencies, including mathematics, if their final ECE centre had been of the highest quality in terms of staff guidance.
In Argentina, Berlinski, Galiani, and Gertler (2006) found bigger gains for children living in poverty. While all children benefited, their gain was 1.6 points higher in mathematics at 3rd grade than the gain of children who were at the country median level of poverty. The curriculum in Argentina was designed to develop communication skills, personal autonomy and behavioural skills, logical and mathematical skills, and emotional skills.
In the U.S., implementation of a conceptually broad mathematics curriculum, involving teachers in professional development and parents having support for learning at home, was associated with benefits for all children. Benefits were greater for children from low-income families (low-income intervention versus comparison group d=0.931; middle-income group versus comparison d=0.723). The rate of change was greater for children from low-income families than for children from middle income families. Effect sizes for change from fall to spring for the low-income group were d>2.0, and for the middle income group d>1.5.
The U.S. Cost, Quality and Outcomes study found the effect of high-quality child care for mathematics skills was greater for children whose mothers had less education than for children whose mothers had more education.
The EPPE study reported that children from some ethnic minority groups (including Black Caribbean and Black African) made greater progress in early number concepts during preschool than white U.K. children or those for whom English was a first language, after taking account of background characteristics. These groups overall had significantly lower cognitive scores at entry to the study in language measures but not nonverbal measures. Sylva et al. (2004) suggest preschool provision may provide opportunity for such children to catch up.
Most studies found no gender differences in gains. However, in the EPPE study, boys gained more than girls for early number concepts if they attended higher-quality centres. The home learning environment scores (measured by parent reports of activities such as reading to the child, playing with letters and numbers, painting and drawing, going to the library) were also lower than for girls, so boys may have been gaining less from home. The study found the home learning environment exerted a significant and independent influence on attainment both at age 3 and the start of primary school. The gains being made by boys from ECE participation were therefore particularly important.
The U.S. Infant Health and Development study reported positive outcomes for the heavier birth weight intervention group during preschool attendance and to age 8 compared with a matched group that did not receive the intervention, but there were no differences in mathematics measures for the lighter birth weight intervention group. The authors suggested the lighter birth weight group may have included a higher proportion of neurologically impaired children who could not benefit from the intervention, who may have needed a more structured and professionally designed home environment, or may have needed continued support beyond the three years of the intervention.
All four studies examining the impact of length of ECE participation and one study on the intensity (hours per week) of ECE participation found gains for mathematics from longer ECE experience. These lasted some time after the ECE experience finished.
The Competent Children, Competent Learners study found that the longer children had attended ECE, the higher their mathematics score at age 12. At age 14 this remained the case controlling for family income, with a significant contrast of around 15 percentage points between those who had attended for 48 months or more, and those who had not. The differences were reduced to the indicative level once maternal qualifications were taken into account. At age 16, though the trend was still evident, it was not statistically significant.
Longer duration in ECE was associated with more progress on number concepts at school entry in the EPPE study: The medium effect sizes increased with ECE duration. (<1 year: d=0.460; 1-2 years: d=0.440; 2-3 years: d=0.568; > 3 years: d=0.631).
One year or two years participation in ECE compared with no ECE participation was associated with development of mathematical competence just after kindergarten (first year of school in the U.S.) entry, with children attending for two years having slightly higher (not statistically significant) scores than those attending for one year in the Barnett and Lamy (2006) study of U.S. pre-K.
A Swedish study (Broberg, Wessels, Lamb and Hwang, 1997) reported higher scores for mathematical ability at age 8½ for children entering centre based care before the age of 40 months compared with those who started after 40 months.
A U.S. randomised trial (Robin et al.., 2006) found children from low-income homes who had been assigned to an extended duration good quality ECE programme (8 hours for 45 weeks per year) improved by 12 standard score points on tests of mathematics skills, while children who had been assigned to participate in a 2½-3-hour programme for 41 weeks per year improved by 7 standard score points, in comparison with children cared for at home or in private care.
ECE centre peer group composition
Two studies have analysed ECE peer group composition in terms of likely advantages from socioeconomic mix, and found that higher socioeconomic mix does benefit children, over and above their own individual family characteristics. In the EPPE and EPPNI studies, children attending ECE centres with a higher proportion of children whose mothers had degrees, higher degrees, or other professional qualifications made more progress in early number concepts at school entry and age 8. The Competent Children, Competent Learners study reported that children whose final ECE centre served mainly middle-class families had higher scores for the cognitive competencies, including mathematics. These associations continued to be statistically significant at age 14 after taking into account family income and maternal qualification, but were not significant at age 16.
Observed measures of the quality of experiences and interactions within the ECE setting have been found to be associated with mathematics development in three longitudinal studies, the Competent Children, Competent Learners study, the English EPPE and Northern Ireland EPPNI studies, and the U.S. Cost, Quality and Child Outcomes study.
The Competent Children, Competent Learners study found patterns of associations with mathematical competency scores and the following items to age 14:
- ECE staff responsiveness to children
- ECE staff guide children in the context of activities
- ECE staff ask open-ended questions
- Children can complete activities
- ECE staff join children in their play
- Provision of a print-saturated environment.
At age 16, some aspects of ECE centre quality were still having an impact over and above that of the equivalent age-5 competency, maternal qualifications, and age-5 family income: ECE staff guiding children accounted for just under 4 percent of the variation in age-16 scores.
In the EPPE study three caregiver interaction scales, which assess negative staff-child interactions (detachment, permissive, and punitive), showed a negative impact on early number concepts progress at the date of starting primary school. Conversely, scoring high on the "Positive relationships scale" was associated with more progress in early number.
Aspects of ECE programme focus had some association with later mathematics competency. Both the EPPE and EPPNI studies used the ECERS-E scale, designed to assess curricular provision in literacy, mathematics, science and environment, and diversity. Neither study found linkages between mathematical competence at age 6½ years and the mathematics subscale, but they found associations with other curriculum subscales. The EPPE study found a statistically significant association between early number concepts and high ratings on the diversity subscale, and a non-significant association with the literacy subscale. The diversity subscale is related to gender equity, multicultural education and "inclusive literacy". The EPPNI study (Melhuish et al.., 2006) found that children who were in ECE centres rated higher on their provision of science attained better scores and made more progress in numeracy.
The Cost, Quality and Child Outcomes study found children in higher-quality care, with higher ratings on observed classroom practices, scored better in math skills after controlling for background variables, than those in lower-quality care. Effect sizes were small: Year 1 child care: d=0.29, Year 2 child care: d=0.28, kindergarten: d=0.20, 2nd grade: d=0.29.
Burchinal, Peisner-Feinberg et al. (2000) pooled data from the Cost, Quality, and Child Outcomes study, North Carolina Head Start Partnership study, and the Public Preschool Evaluation project. All these studies included the same measure of child care quality and family selection factors, and similar or the same measures of language and pre-academic development and social skills. In the short term, children in poor quality centres had significantly lower mathematics scores than those in high-quality (medium effect size d=0.48) and lower mathematics scores (not statistically significant, effect size d=0.33) than those in medium quality.
Structural aspects of staff: child ratios (more adults to children) and smaller group size were associated with enhanced mathematical abilities in a Swedish study (Broberg et al.., 1997) at age 8. Those in centre-based ECE did better than those in family day care.
Early et al. (2006) found teachers' education (years of education, highest degree, and Bachelor's degree versus no Bachelor's degree) linked to gains on standardised measures of mathematics skills across the pre-K year. Children gained additional points where teachers had a Bachelor's degree compared with less than a Bachelor's degree. These associations were not found with language and early literacy. The authors suggested that because of a national focus in the U.S. on language and early literacy, it may be that teacher education programmes were preparing teacher trainees to work in that area, but were not specifically working with them on mathematical thinking.
In the EPPE study, higher levels of teacher qualifications were associated with positive aspects of adult-child interaction, which in turn benefited child outcomes, i.e. the linkage with outcomes was indirect through the association of qualifications with better quality ECE provision.
Starkey et al. (2004) examined the implementation of a conceptually broad mathematics pre-kindergarten curriculum, in which teachers undertook professional development and changed their curriculum, and parents were offered materials and curriculum guide sheets and classes to help them support mathematics learning at home. Over the period of a year, intervention children in comparison with control children gained significantly enhanced mathematical knowledge (low-income intervention versus comparison group d=0.931; middle-income group versus comparison d=0.723)8
Reading and literacy
Positive impacts of ECE participation on reading and literacy were reported in 13 studies, where small to medium effect sizes were maintained through the early years of schooling, declining over the long term. Mixed impact (neutral for one group, negative for another) was found in one study. No impact was found in three studies using survey data.
|Intervention studies||Consistent benefits in 3 studies (small to large effect sizes d=0.34-0.89)||Consistent benefits in 3 studies (small to medium effect sizes d=0.17-0.44)|
|Everyday ECE Studies||Benefits in 8 studies, mixed impact in 1 study||Consistent benefits in 1 study (summary of 13 pre-K evaluations)|
|Survey studies||No significant effects in 3 studies (but positive correlation in 2)|
The U.S. intervention studies reported medium to large effects for the full intervention versus no intervention. At age 8, these were 0.34 for the Chicago Child Parent-Centre programme and 0.89 for the Abecedarian programme for teaching achievement.
Studies in the U.S., U.K., Sweden, and Bangladesh that followed children through from ECE programmes to the first and second year of school reported small to large effects, ranging from d=0.20 for a programme in rural Bangladesh to f=0.68 for Andersson's (1989) Swedish study. The large effect size in Andersson's study was at age 8 for children entering ECE between birth and age 1. The EPPE study found a medium effect size for language (d=0.44), and a smaller effect size for pre-reading (d=0.28) at age 6. Sammons et al. (2002) noted that "after controlling for the child, parent and home learning environments factors, a child with preschool experience attains on average a pre-reading score of 2.7 points higher than a child without such experiences" (p. 53). Although in the EPPE study, the effect sizes for reading are smaller than those for mathematics, there was less decline in effect sizes on reading tests over the first two years of school than for mathematics attainment (shift from approximately d=0.28 at school entry, to d=0.23 at age 6, to d=0.28 at age 7).
In their synthesis of early Head Start evaluations, Love et al. (2005) found overall positive impacts on language measures at age 3 (small effect sizes ranging from d=0.09 for centre-based programmes to d=0.23 for combined centre-based/home support programmes). They found significant reduction in those scoring more than 1 standard deviation below the mean on the PPVT-111 (receptive vocabulary) (small effect sizes d=-0.05 to d=-0.23).
In Argentina, Berlinski, Galiani, and Gertler (2006) found positive impacts of pre-primary education on 3rd grade standardised Spanish tests, following Argentina's expansion of good quality universal pre-primary education between 1993 and 1999 (small effect size d=0.23).
The effects of ECE participation on reading and literacy were reported for five long-term studies, with all showing positive impacts to age 14 or later.
The largest effect sizes for reading were from the Abecedarian study (d=0.89 at age 8, d=0.48 at age 12, d=0.44 at age 15). The Chicago Child Parent-Centre programme found a medium effect size of d=0.34 at 3rd grade, and small effect sizes of d=0.17 at fifth grade and d=0.17 at eighth grade (Karoly et al.., 2005). The High/Scope Perry Preschool study showed the programme group compared with a control group had significantly higher scores in general literacy at age 19.
Gilliam and Zigler's (2004) summary of 13 U.S. pre-K evaluations reported that in Michigan 24 percent more pre-kindergarten participants passed the school administered reading test at 4th grade (the final level assessed). New York-EFK found impacts at 6th grade (the final level assessed). Maryland found statistically significant impacts at 5th, 8th, and 9th grade but not at 10th grade.
The U.K. National Evaluation of Sure Start Local Programmes (SSLPs) found no impact on verbal abilities for more advantaged families (in a relatively disadvantaged community) from living in a SSLP community and adverse effects for the most disadvantaged families with 3-year-olds. Children of families with teen mothers, lone parents, and families with no unemployed parents had lower levels of verbal skills when living in a SSLP community than those who were in Sure Start-to-be communities. These programmes had core services of family/parent support, child and maternal health, and play and child care, and were intended to support social inclusion of very disadvantaged families. Quality and type of ECE services varied and it was not clear whether the teen families availed themselves of the services, so the impact of ECE per se was not separable.
Studies using U.S., Dutch, and Canadian national survey data reported no significant effects of ECE participation (the U.S. study at age 6; the Canadian study at age 7; and the Dutch study at grade 4). There were weak positive relationships between ECE participation and language competencies in the analysis of Canadian and Dutch survey data, but these disappeared when background characteristics were taken into account.
Differences for Population Groups
Language and literacy seems to be an aspect where children who start at lower levels are particularly helped by ECE participation to catch up. Data from seven studies examined differential effects of ECE participation for population groups. In these studies, good quality ECE enabled ethnic minority children, children with English as an Additional Language (EAL), and children from low-income families to develop at a faster rate.
Barnett et al.., (2005) analysis of pre-K in five states found some evidence of a more positive effect of pre-K participation on early print awareness for children in low-income families. The overall gain for those who qualified for a free/partially subsidised school lunch was about three more items correct than for those who did not qualify. In Oklahoma and Carolina extra gains for children from low-income homes were about 8 percent more items correct. In Argentina, Berlinski, Galiani, and Gertler (2006) found bigger gains for children living in poverty (3.2 points higher in Spanish), compared with those at the median level of poverty.
Burchinal, Peisner-Feinberg et al.'s (2000) secondary analysis of data from the Cost, Quality and Child Outcomes study, North Carolina Head Start Partnership study, and the Public Preschool Evaluation project found a greater positive impact of good quality of the ECE setting for language skills of "children of colour" compared with white children. Among "children of colour", the large effect size for differences between the adjusted mean for children in poor quality and those in medium quality was d=1.12, and those in high-quality was d=1.54. Among white children, the effect sizes for differences were medium: d=0.26 and d=0.48 respectively.
The EPPE study also found differences related to ethnicity. Children from Black Caribbean, Black African, Black other, Indian, and mixed heritage ethnic groups made more progress on pre-reading than the White U.K. ethnic group, after controlling for SES and maternal education level. They started preschool with lower prior attainment and made significant gains during preschool. EAL children who had significantly lower scores at entry also made more progress than children whose first language was English during preschool.
The Cost, Quality and Child Outcomes study found (Burchinal, Roberts et al.., 2000) that centres that met professional recommendations regarding teacher education tended to have girls with more enhanced receptive language skills.
The U.S. Infant Health and Development study found positive outcomes for the heavier birth weight group in reading over the preschool years and at age 8 compared with a matched group that did not receive the intervention, but there were no differences in reading measures for the lighter birth weight intervention group.
Consistent findings in five studies linked longer ECE duration with reading and literacy gains. Benefits of longer duration were found in both services targeting children from low-income families and those catering for children from a wide socioeconomic range. The advantages of longer duration diminished over time.
Starting ECE before age 3 was associated with reading and literacy measures at school entry (English EPPE project), at 8½ years (Swedish study—Broberg, Wessels, Lamb & Hwang (1997) and at age 10 (New Zealand Competent Children, Competent Learners study). The overall length of ECE experience in the Competent Children, Competent Learners study did not make an independent contribution at age 14. These studies included children from a range of family backgrounds.
Two years' ECE experience versus one year. Two U.S. studies (Barnett & Lamy, 2006; Reynolds, 1995) of children from low-income families attending good quality ECE found some literacy measures at school entry favoured children with two years' duration. The differences between the groups at school entry did not last.
Full-time versus part-time
Findings about intensity of duration are mixed. One U.S. study, Robin et al. (2006), found children from low-income families in good quality extended duration programmes (8 hours for 45 weeks per year) improved 11 standard score points on vocabulary compared with children in half-day programmes (6 standard score points) at the end of first grade. In Loeb et al.'s (2005) U.S. study, children attending for 15 to 30, and over 30 hours, scored approximately 8 percent of a standard deviation higher on pre-reading skills than those attending for fewer than 15 hours per week.
The English EPPE study of everyday ECE serving children from a range of family backgrounds found no evidence that full-time provision resulted in better outcomes than part-time.
ECE centre peer group composition
The EPPE study found:
- Children attending centres, where there was a higher proportion of children with below average attainment, made less progress in pre-reading.
- Children attending ECE centres, where there were a higher proportion of children whose mothers had degrees, higher degrees, or other professional qualifications, made more progress in pre-reading and language. The Northern Ireland EPPNI study made a similar finding.
The Competent Children, Competent Learners study found that children whose final ECE centre served mainly middle-class families had higher cognitive scores for the cognitive competencies, including reading. These associations continued to be significant at age 14 after taking into account family income and maternal qualification, but were not significant at age 16.
Positive associations between children's literacy performance and aspects of quality ECE provision have been reported in studies in many countries, except for one U.S. study.
In the Competent Children, Competent Learners study patterns of associations with reading and literacy measures and the following quality items were found to age 14:
- ECE staff responsiveness to children
- ECE staff guide children in the context of activities
- ECE staff ask open-ended questions
- ECE staff join children in their play
- Provision of a print-saturated environment.
By age 16, there were no longer statistically significant associations between these items and literacy, once the age-5 score and social characteristics had been included in the model.
Responsive and challenging adult-child relationships were also associated with pre-reading levels in the EPPE study, and a negative effect at school entry was found for children who had attended centres scoring highly on negative staff-child interactions (detachment, permissive, and punitive). The EPPE study also found a statistically significant association between pre-reading levels at school entry and high ratings on the ECERS-E literacy subscale.
The U.S. Cost, Quality, and Child Outcomes study (Peisner-Feinberg et al.., 1999) calculated effect sizes for child care quality aspects to grade two. Children attending child care with higher ratings on observed classroom practices and teacher-child closeness scored better than those in lower-quality care for receptive language ability, after controlling for background variables. The differences were between the lowest quartile on the quality measure and the highest quartile after adjusting for background variables and the other child care quality measures. Unlike the associations between receptive language ability and the classroom practices index, the association with teacher ratings of teacher-child closeness did not decline over time. Effect sizes were as follows:
- observed classroom practices: child care year 1, d=0.60; child care year 2, d=0.51; kindergarten, d=0.30, second grade, d=0.14
- teacher-child closeness: child care year 1, d=0.30; child care year 2, d=0.35; kindergarten, d=0.17, second grade, d=0.33.
A study using pooled data from the Cost, Quality, and Child Outcomes study, North Carolina Head Start Partnership study, and the Public Preschool Evaluation project (Burchinal, Peisner-Feinberg et al.., 2000) found children in poor quality centres had significantly lower reading scores than those in medium quality (medium effect size d=0.42) or high-quality (medium effect size d=0.52).
Montie, Xiang, and Schweinhart (2006) analysed data from 10 countries in the IEA Pre-primary Project (Finland, Greece, Hong Kong, Indonesia, Ireland, Italy, Poland, Spain, Thailand, and United States) to identify how process and structural characteristics of the ECE settings children attended at age 4 are related to age-7 competencies. In all countries children in ECE settings with free choice activities (teachers allow children to choose their own activities) achieved significantly higher average language scores at age 7 than their counterparts in centres where personal care and group activities predominated, and a nearly significant higher score than counterparts in centres where pre-academic activities predominated. The authors suggested free choice activities may be more interesting and engaging to the child, and the difficulty level more suitable than those that are proposed by teachers. In addition, these activities allow opportunities for children to interact verbally with other children, and for teachers to engage in relevant conversation and introduce new vocabulary.
Some findings varied across countries, and these seemed to relate to differences in countries' culture and beliefs about children:
- Increased adult-child interaction was related to better age-7 language scores in countries that have less adult-centred teaching or activities that require group responses, and poorer language scores in countries that have more adult-centred teaching or activities that require group responses. The authors suggested that in countries where child-centred teaching is typical and children are encouraged to express their views, "adult-child interaction is likely to encourage independent thought and freedom of expression, thus fostering language learning" (Montie et al.., 2006). Conversely where adult-centred teaching is the norm, children are expected to listen, learn from, and obey teachers. In these situations, with increased adult-child interaction, children may have less opportunity to plan their play or solve problems.
- Increased child-child interaction was related to better age-7 language scores in countries that have fewer whole-group activities or more teachers who rank language skills among the most important, and poorer language scores in countries that have more whole-group activities or fewer teachers who rank language skills among the most important.
Votruba-Drzal et al.'s (2004) study of U.S. low-income families found no significant associations between child care quality and development of quantitative and reading skills. The authors suggest two reasons for their findings being different from other studies: low-income children need higher-quality ECE than was found in the study; and they may also need longer consistent experience to gain cognitively.
Structural measures of quality
Positive associations with structural measures of quality were reported in four studies.
Teacher Education and Ratios:
Positive associations with levels of ECE teacher education were consistently found —generally higher levels of teacher education were associated with children's reading and language progress in the first two years of schooling. These included Montie et al.'s (2006) findings from the IEA Pre-primary Project in 10 countries that as the level of teacher education increased, children's age-7 language performance improved.
The NICHD Early Child Care Research Network (1999) also found linear associations between the number of recommended standards9 for quality (teacher training, teacher education, group size, and teacher: child ratios) met and language comprehension scores at 36 months. There was no evidence of threshold effects. Not meeting any of the quality standards was related to lower than average scores at 36 months for language comprehension, and meeting all of them with above average scores. Child outcomes were best predicted by staff: child ratio at 24 months and caregiver training and education at 36 months. These authors also noted that classes for older children were more likely to meet the standards than classes for infants and toddlers.
Burchinal, Roberts et al.'s (2000) study of 89 African American children attending community based child care centres (they enrolled before age 12 months) found that children in classrooms that met professional recommendations regarding staff: child ratios scored significantly higher across time for receptive communication with an estimated difference of 1.01 points in terms of developmental months at each age—12, 18, 24, and 36 months (effect size d=0.34), and higher expressive communication skills at 36 months. They also had significantly higher overall communication skills with adjusted means of 103.8 for children in classrooms with good ratios and 98.1 for children in classrooms with poor ratios (effect size d=0.54). These are large effect sizes given the sample in this study was small. The authors suggested this finding is consistent with the fact that scaffolded conversations are especially important for language development during early childhood.
They also found girls (but not boys) in classes with a lead teacher who had over 14 years' education showed larger gains in receptive and expressive language over time. This may have been because girls were more likely to seek out the attention of adults and converse with them.
In the EPPE study, percentage of time qualified staff spent working with children was also related to pre-reading progress at school entry. Montie et al. (2006) suggested teachers with more education use more words and more complex language when communicating with children.
Montie et al. (2006) found that group size did not relate to children's age-7 language scores for the 10 countries studied. The NICHD study above also did not find that group size predicted child outcomes at 36 months when considered separately from other structural measures. Montie et al. suggested group size could be country specific, giving as an example that having a large number of children in the classroom is considered desirable in Japan because children have more opportunity to learn from one another and it reduces the demand for one-to-one interaction with teachers. This is in contrast to U.S. studies where group size is shown to be associated with positive child outcomes.
School performance, cognitive scores, and IQ
In this section, studies reporting on knowledge-based competencies of school performance, school achievement tests, grade retention, and special education placement, or overall cognitive scores and IQ are examined.
The three U.S. intervention studies, Abecedarian, Chicago Child-Parent Centre, and High/Scope Perry Preschool study, reported strong positive impacts of ECE participation on cognitive competence through to school leaving age. Medium to large effect sizes were found for school performance (0.33 to 0.77), IQ (0.5 to 0.9 at school entry), fewer grade retentions (2 percent to 23 percent at age 15), and reductions in special education placements (23 percent to 48 percent).
General Studies and Survey Data:
Positive effects of ECE participation on cognitive outcomes were found in most studies, overall. No negative effects were found. The only studies reporting no impact on any of the measures were in the U.S.
(ECE versus none)
|Highest grade, school leaving, higher level school|
|Consistent benefits in nine studies in Bangladesh, Sweden, U.S., Northern Ireland, and England||
Consistent benefits in 2 U.S.,|
and 1 Canadian study
|Consistently reduced grade retentions in 3 U.S. studies||Lower rates of special education placement in 2 U.S. studies, no difference in 1 U.S. study||Benefits in 1 Uruguayan study, 1 U.S. study, 1 West Germany study|
Cognitive performance effect sizes close to school entry, where reported, ranged from medium (d=0.39) in Oklahoma's pre-K programme (Gormleyet al., 2005) to large (f=1.0) in rural Bangladesh (Aboud, 2006) where children in villages with preschools were compared with children in villages without preschools.
Love et al. (2005), in an evaluation of early Head Start, reported small effect sizes from a combined ECE and parent support programme of d=0.28 on Bayley MDI scores and d=0.34 on percentage of children scoring below 85 on PPVT-IIII when children were aged 3.
Effect sizes increased in a U.S. study (Bagnato et al., 2002) following children enrolled in a high-quality early childhood initiative over the time of preschool attendance. The impact on a composite score of teacher and parent assessed developmental and behavioural outcomes increased from -.0084 (50th percentile) to .8489 (80th percentile) after 12 months.
Two studies found gains in the early years magnified as children grew older. In Andersson's (1992) Swedish study, children who entered ECE between birth and 1 year compared with home care children had significantly better teacher assessed school performance at age 8 (medium effect size d=0.49) and age 13 (large effect size d=0.74). Berlinski, Galiani, and Manacorda (2006), using Uruguayan household survey data, found significant positive effects for those with pre-primary education compared with those without on number of years schooling completed. At age 10, those with pre-primary education had 0.28 years more completed years of schooling than those without, and at age 12, 0.32 more years. By 13, these children were less likely to drop out of school, and by 16, they were 27 percentage points more likely to be in school and to have accumulated more than one year of extra education.
On the other hand, the EPPE study (Sammons et al., 2002) found effects of ECE versus none on nonverbal reasoning and spatial awareness/reasoning were stronger at age 6 than at age 7. The authors noted that this could be accounted for by the change in the measures used in the study from standardised tests (British Ability Scales) at school entry and age 6 to national assessments at age 7; or by the "rise of the primary effect", i.e. the result of the accumulating and powerful effects of the primary school (Sylva et al., 2004, p. 44).
Differences for population groups
Good quality ECE had benefits for children with learning or behavioural difficulties in two studies. Those who were at risk of these difficulties benefited significantly from good quality provision and showed further benefit when there was a mixture of children from different social backgrounds (EPPE study). The EPPE study found that more of the children who did not have ECE participation were "at risk" of special educational needs at primary school entry, and were identified by teachers as showing some form of special educational needs during the early years of schooling. Bagnato et al. (2002), evaluating a high-quality Early Childhood Initiative in Pittsburgh, found progress of 1.6 months of gain for every month of programme participation for children with mild developmental delays. Approximately 14 children at the start demonstrated delays that would have categorised them with a mental health diagnosis. At the end of two years, only one showed significant needs.
A second U.S. study (Booth & Kelly, 2002), where quality of setting was not assessed, found no effects of child care participation on cognitive development of children with special needs compared with home children at 30 months.
Evidence was consistent across all nine studies investigating quality of associations between good quality ECE and cognitive development.
Those investigating structural features found relationships between levels of teacher education and staff: child ratios and cognitive development. The NICHD Early Child Care Research Network found linear associations between the number of recommended standards10 for quality (teacher training and teacher: adult ratios) met and child outcomes at 24 and 36 months, with higher effect sizes at 36 months.
Montie et al.'s (2006) analysis across findings from the IEA Pre-primary Project for 10 countries found that less time spent in whole-group activities was associated with better age-7 cognitive performance. It did not find associations with group size. This study also found that as the number and variety of materials in settings increased, children's age-7 cognitive performance improved. Increased adult-child interaction was related to better age-7 cognitive performance in countries where teachers included a lot of free choice activities, and poorer cognitive performance in countries where teachers proposed few free choice activities.
The Cost, Quality and Child Outcomes study (Helburn, 1995; Peisner-Feinberg et al., 1999) and Smith's (1996) New Zealand study of infant child care centres found high teacher compensation was linked to higher ratings of the quality of the ECE service, which in the Cost, Quality and Child Outcomes study was linked to child outcomes.
Marcon (2002), in a U.S. study to compare different preschool models (child-initiated, academically directed, or a combination approach) on later school performance, found children whose preschool class had been academically directed (and so children had less choice and were not enabled to investigate and think for themselves) had significantly lower grades in year six than children whose preschool class had been child-initiated (small effect size d=0.34). On the other hand, those whose preschool was academically directed were retained in grade less often at end of the 5th year, perhaps because the academic model was closer to the school model.
The EPPE study found that integrated ECE centres that included flexible hours for child care and health and family support services, and nursery schools had higher scores on ECE quality and better cognitive outcomes than playgroups, private day nurseries, and local authority day nurseries. The integrated centres and nursery schools have a higher proportion of trained teachers than the others.
The REPEY study (Researching Effective Pedagogy in the Early Years) followed on from the EPPI and EPPNI studies. It analysed the pedagogic models and practices being applied by 12 settings classified in the EPPE study as having good to excellent practice in terms of the children's developmental progress in cognitive, social, or dispositional outcomes. Good outcomes in terms of cognitive, social, and dispositional outcomes for children were linked to early years settings that:
- view cognitive and social development of children as complementary and do not prioritise one over the other
- have strong leadership and long-serving staff (three years plus)
- provide a strong educational focus with trained teachers working alongside and supporting less qualified staff
- provide children with a mixture of practitioner initiated group work and learning through freely chosen play
- provide adult-child interactions that involve 'sustained shared thinking' and open-ended questioning to extend children's thinking
- have practitioners with good curriculum knowledge and knowledge and understanding of how young children learn
- have strong parent involvement, especially in terms of shared educational aims with parents
- provide formative feedback to children during activities and provide regular reporting and discussion with parents about their child's progress
- ensure behaviour policies in which staff support children in rationalising and talking through their conflicts
- provide differentiated learning opportunities that meet the needs of particular individuals and groups of children (Sylva et al., 2004, p. i).
Their analysis showed an association between curriculum differentiation and matching in terms of cognitive challenge, and "sustained shared thinking". In respect to parent involvement, children had better cognitive outcomes in those settings that encouraged continuity of learning between the early years setting and home, through sharing educational aims with parents, engaging parents in regular ongoing assessment of children's learning, and supporting parents where this support was combined with educational aims. The qualitative evidence also suggested that the better a setting did on each of these elements of pedagogic practice, the greater was the positive effect on children's cognitive progress (Siraj-Blatchford, 2004; Siraj-Blatchford & Sylva, 2004).
Positive associations with length of ECE experience have been reported. Generally, children who have attended ECE for longer show higher cognitive performance levels, provided that the ECE is good quality. Using data from the Infant Health and Development Program, Lee (2005) found that hours spent in the good quality care provided by the intervention was positively related to cognitive outcomes at age 3. For children who were not in the intervention, hours spent in care (which was likely of poor quality) was negatively related to cognitive performance.
There is some indication that longer duration may impact differentially on children from lower- and higher-income families. Loeb et al.'s (2005) analysis of data from a large nationally representative U.S. sample (14,162) of kindergarteners estimating the influence of different amounts of participation in preschool centres on cognitive and social-emotional outcomes found, on average, that children attending centres for 15 to 30 hours per week experienced stronger cognitive gains than those attending for less than 15 hours per week. Attendance for more than 30 hours per week did not yield additional gains on average. For children from lower-income families, additional hours (more than 30) did advance cognitive gains, but for children from higher-income families no further gains were found.
Consistent evidence from a large body of international and New Zealand evidence shows that ECE participation is positively associated with gains in mathematics and literacy, and other measures of cognitive performance such as intelligence tests, school readiness, grade retention, and special education placement or identification as having special educational needs, after controlling for home environment and other background variables. Short-term, medium to large effect sizes on all the outcome measures were reported in U.S. intervention studies targeting children from low-income families, and combining good quality ECE with parenting support/education, and small to medium effect sizes from ECE participation were found in studies reporting on everyday ECE experiences. Many of these were general population studies. Long term, participation in high-quality ECE was linked to gains on these outcomes in most studies.
Effect sizes tended to diminish during schooling, but not in all studies. Reasons for different trajectories are complex, but students' early learning as well as subsequent schooling and other experiences have a powerful effect. It is notable that impacts endured long term.
Aspects of ECE that can affect ECE impact
Children in high-quality ECE settings experienced significantly greater cognitive gains than children in low-quality settings, with medium effect sizes reported from participation in high- compared with low-quality services in mathematics, reading/literacy, and school performance. In evaluating quality, measures of process, structure, curriculum, and relationships with parents have been used.
Positive effects of ECE participation were found in settings described as good quality in terms of adult-child interactions that are responsive, cognitively challenging, and encourage joint attention and negotiation or "sustained shared thinking".
There is evidence that a curriculum where children can investigate and think for themselves is associated with better cognitive performance in later schooling than one that is academically oriented.
Significant associations were found between staff: child ratios, teacher qualifications and education, teacher compensation, and children's cognitive outcomes. Group size was not measured in most studies, but less time (but not small group size) spent in whole-group activities was associated with better age-7 cognitive performance in analysis of the IEA Pre-primary Project. It may be that how children are grouped within settings, rather than child-initiated overall centre size, is what counts for quality.
This clear evidence showing associations between cognitively challenging adult-child interactions and gains for children, and the association between teacher qualifications and these kinds of interactions suggests a mediated path may be through teacher qualifications (teachers drawing on knowledge and experience of pedagogy), to ways in which teachers interact with children, to child outcomes. One study (NICHD ECCRN, 2002), using structural equation modelling, found a mediated path from structural indicators of quality (teacher qualifications and staff: child ratios) through process quality to child outcomes, including cognitive competence. These authors suggest that "more caregiver training may lead to better interactions between children and adults, while lower ratios may lead to more interactions" (p. 206).
The EPPE study found that having strong parental involvement in terms of educational aims could help improve the home learning environment and reinforce learning between home and the ECE service.
Longer duration of ECE experience is linked with cognitive gains for children from a range of family socioeconomic backgrounds. Benefits of longer duration diminish over time, but may still be evident for mathematics and other schooling outcomes long term. High-quality ECE with longer duration has the strongest effects. An early starting age before age 3 is associated with gains, but there is mixed evidence about whether starting before age 2 is more advantageous than starting between age 2 and 3.
Hours per week
Full-time attendance has no benefits for cognitive outcomes over part-time attendance in studies of children from a range of socioeconomic backgrounds. Some U.S. studies found children from low-income homes attending good quality ECE services gained more from full-time attendance (more than 30 hours) in literacy, mathematics, and other cognitive outcomes than those with part-time attendance.
New Zealand and international studies found gains for children from low-income/disadvantaged homes could be greater than for most other children in mathematics and literacy, if their ECE centre was of high-quality.
English as an Additional Language:
Children for whom English is an additional language, and children from some ethnic minority groups, made greater progress on early number concepts and literacy/language measures during ECE participation than others in the English EPPE study. These children tended to start with lower scores on language measures (but not nonverbal) and the ECE experience helped them start to catch up with peers.
Gender differences were found in three studies:
- Boys gained more than girls on early number concepts over the time of ECE attendance in the English EPPE study. They also had lower home learning environment scores (measured by parent reports of activities such as playing with letters and numbers, going to the library, reading to the child, teaching songs and nursery rhymes, painting and drawing, teaching the alphabet and numbers, taking children on visits, and creating regular opportunities for them to play with their friends at school) than girls, so ECE could be more effective.
- In the U.S. Cost, Quality and Child Outcomes study, centres that met professional recommendations regarding teacher education tended to have girls with more enhanced receptive language skills than boys.
Socioeconomic Mix of ECE Centre:
Children attending ECE centres with a middle class/better maternally educated mix had greater gains for mathematics, literacy, and other cognitive outcomes, both short term and long term.
- A New Zealand study (Young-Loveridge, Carr, & Peters, 1995) also showed how professional development with teachers focused on mathematics contributed to enhanced mathematical experiences and resources within the ECE centres.
- Ratios: 1:3 at 6 and 15 months, 1:4 at 24 months, 1:7 at 36 months.
Group size: 6 at 6 and 15 months, 8 at 24 months, 14 at 36 months.
Qualifications: formal post-high training in child development, ECE, or related field at all four ages.
- Ratios: 1:3 at 6 and 15 months, 1:4 at 24 months, 1:7 at 36 months.
Group size: 6 at 6 and 15 months, 8 at 24 months, 14 at 36 months.
Qualifications: formal post-high training in child development, ECE, or related field at all four ages.